mirror of
https://github.com/MODSetter/SurfSense.git
synced 2026-05-06 06:12:40 +02:00
Remove unwired multi_agent_chat package.
This commit is contained in:
CREDO23
parent
216a678f1a
commit
d675d4df3f
93 changed files with 0 additions and 3697 deletions
|
|
@ -1,132 +0,0 @@
|
|||
"""
|
||||
Multi-agent chat (LangChain Subagents pattern).
|
||||
|
||||
**Layout (SRP)**
|
||||
|
||||
- :mod:`expert_agent.builtins` — general categories from the tool registry (research, memory, deliverables — not tied to one vendor).
|
||||
- :mod:`expert_agent.connectors` — external integrations (one subgraph per product where split).
|
||||
- :mod:`core` — prompts, compiled subgraph helper, delegation, registry subsets, tool-factory kwargs (:mod:`core.bindings`).
|
||||
- :mod:`routing` — supervisor-facing ``@tool`` routers → domain invoke.
|
||||
- :mod:`supervisor` — orchestrator graph + ``supervisor_prompt.md``.
|
||||
- :mod:`integration` — async ``create_multi_agent_chat`` composer (partitions MCP tools into experts).
|
||||
|
||||
Documentation:
|
||||
https://docs.langchain.com/oss/python/langchain/multi-agent
|
||||
https://docs.langchain.com/oss/python/langchain/multi-agent/subagents
|
||||
|
||||
Display name: ``multi-agent-chat`` — Python package: ``multi_agent_chat``.
|
||||
"""
|
||||
|
||||
from app.agents.multi_agent_chat.expert_agent.builtins.deliverables import (
|
||||
build_deliverables_tools,
|
||||
build_deliverables_domain_agent,
|
||||
)
|
||||
from app.agents.multi_agent_chat.expert_agent.builtins.memory import (
|
||||
build_memory_tools,
|
||||
build_memory_domain_agent,
|
||||
)
|
||||
from app.agents.multi_agent_chat.expert_agent.builtins.research import (
|
||||
build_research_tools,
|
||||
build_research_domain_agent,
|
||||
)
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.calendar import (
|
||||
build_calendar_domain_agent,
|
||||
build_calendar_tools,
|
||||
)
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.confluence import (
|
||||
build_confluence_tools,
|
||||
build_confluence_domain_agent,
|
||||
)
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.discord import (
|
||||
build_discord_tools,
|
||||
build_discord_domain_agent,
|
||||
)
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.dropbox import (
|
||||
build_dropbox_tools,
|
||||
build_dropbox_domain_agent,
|
||||
)
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.gmail import (
|
||||
build_gmail_tools,
|
||||
build_gmail_domain_agent,
|
||||
)
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.google_drive import (
|
||||
build_google_drive_tools,
|
||||
build_google_drive_domain_agent,
|
||||
)
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.luma import (
|
||||
build_luma_tools,
|
||||
build_luma_domain_agent,
|
||||
)
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.notion import (
|
||||
build_notion_tools,
|
||||
build_notion_domain_agent,
|
||||
)
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.onedrive import (
|
||||
build_onedrive_tools,
|
||||
build_onedrive_domain_agent,
|
||||
)
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.teams import (
|
||||
build_teams_tools,
|
||||
build_teams_domain_agent,
|
||||
)
|
||||
from app.agents.multi_agent_chat.core import (
|
||||
REGISTRY_ROUTING_CATEGORY_KEYS,
|
||||
TOOL_NAMES_BY_CATEGORY,
|
||||
build_domain_agent,
|
||||
build_registry_dependencies,
|
||||
build_registry_tools_for_category,
|
||||
compose_child_task,
|
||||
connector_binding,
|
||||
extract_last_assistant_text,
|
||||
read_prompt_md,
|
||||
)
|
||||
from app.agents.multi_agent_chat.integration import create_multi_agent_chat
|
||||
from app.agents.multi_agent_chat.routing import (
|
||||
DomainRoutingSpec,
|
||||
build_supervisor_routing_tools,
|
||||
routing_tools_from_specs,
|
||||
)
|
||||
from app.agents.multi_agent_chat.supervisor import build_supervisor_agent
|
||||
|
||||
__all__ = [
|
||||
"REGISTRY_ROUTING_CATEGORY_KEYS",
|
||||
"TOOL_NAMES_BY_CATEGORY",
|
||||
"DomainRoutingSpec",
|
||||
"build_calendar_domain_agent",
|
||||
"build_confluence_tools",
|
||||
"build_confluence_domain_agent",
|
||||
"build_deliverables_tools",
|
||||
"build_deliverables_domain_agent",
|
||||
"build_discord_tools",
|
||||
"build_discord_domain_agent",
|
||||
"build_domain_agent",
|
||||
"build_dropbox_tools",
|
||||
"build_dropbox_domain_agent",
|
||||
"build_gmail_tools",
|
||||
"build_gmail_domain_agent",
|
||||
"build_calendar_tools",
|
||||
"build_google_drive_tools",
|
||||
"build_google_drive_domain_agent",
|
||||
"build_luma_tools",
|
||||
"build_luma_domain_agent",
|
||||
"build_memory_tools",
|
||||
"build_memory_domain_agent",
|
||||
"build_notion_tools",
|
||||
"build_notion_domain_agent",
|
||||
"build_onedrive_tools",
|
||||
"build_onedrive_domain_agent",
|
||||
"build_registry_dependencies",
|
||||
"build_registry_tools_for_category",
|
||||
"build_research_tools",
|
||||
"build_research_domain_agent",
|
||||
"build_supervisor_agent",
|
||||
"build_supervisor_routing_tools",
|
||||
"build_teams_tools",
|
||||
"build_teams_domain_agent",
|
||||
"connector_binding",
|
||||
"compose_child_task",
|
||||
"create_multi_agent_chat",
|
||||
"extract_last_assistant_text",
|
||||
"read_prompt_md",
|
||||
"routing_tools_from_specs",
|
||||
]
|
||||
|
|
@ -1,25 +0,0 @@
|
|||
"""Cross-cutting building blocks (prompts, agents, delegation, registry) — not domain logic."""
|
||||
|
||||
from app.agents.multi_agent_chat.core.agents import build_domain_agent
|
||||
from app.agents.multi_agent_chat.core.bindings import connector_binding
|
||||
from app.agents.multi_agent_chat.core.delegation import compose_child_task
|
||||
from app.agents.multi_agent_chat.core.invocation import extract_last_assistant_text
|
||||
from app.agents.multi_agent_chat.core.prompts import read_prompt_md
|
||||
from app.agents.multi_agent_chat.core.registry import (
|
||||
REGISTRY_ROUTING_CATEGORY_KEYS,
|
||||
TOOL_NAMES_BY_CATEGORY,
|
||||
build_registry_dependencies,
|
||||
build_registry_tools_for_category,
|
||||
)
|
||||
|
||||
__all__ = [
|
||||
"REGISTRY_ROUTING_CATEGORY_KEYS",
|
||||
"TOOL_NAMES_BY_CATEGORY",
|
||||
"build_domain_agent",
|
||||
"build_registry_dependencies",
|
||||
"build_registry_tools_for_category",
|
||||
"compose_child_task",
|
||||
"connector_binding",
|
||||
"extract_last_assistant_text",
|
||||
"read_prompt_md",
|
||||
]
|
||||
|
|
@ -1,5 +0,0 @@
|
|||
"""Compiled subgraph factories shared by domain slices."""
|
||||
|
||||
from app.agents.multi_agent_chat.core.agents.domain_graph import build_domain_agent
|
||||
|
||||
__all__ = ["build_domain_agent"]
|
||||
|
|
@ -1,27 +0,0 @@
|
|||
"""Compile a domain LangGraph agent from a co-located prompt + tool list."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from collections.abc import Sequence
|
||||
|
||||
from langchain.agents import create_agent
|
||||
from langchain_core.language_models import BaseChatModel
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.prompts import read_prompt_md
|
||||
|
||||
|
||||
def build_domain_agent(
|
||||
llm: BaseChatModel,
|
||||
tools: Sequence[BaseTool],
|
||||
*,
|
||||
prompt_package: str,
|
||||
prompt_stem: str = "domain_prompt",
|
||||
):
|
||||
"""``create_agent`` + ``{prompt_stem}.md`` loaded from ``prompt_package``."""
|
||||
system_prompt = read_prompt_md(prompt_package, prompt_stem)
|
||||
return create_agent(
|
||||
llm,
|
||||
system_prompt=system_prompt,
|
||||
tools=list(tools),
|
||||
)
|
||||
|
|
@ -1,5 +0,0 @@
|
|||
"""Search-space / DB kwargs shared by main-chat tool factories (distinct from ``expert_agent.connectors`` integrations)."""
|
||||
|
||||
from app.agents.multi_agent_chat.core.bindings.binding import connector_binding
|
||||
|
||||
__all__ = ["connector_binding"]
|
||||
|
|
@ -1,18 +0,0 @@
|
|||
"""Shared kwargs dict for main-chat tool factories (DB session + search space + user)."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from sqlalchemy.ext.asyncio import AsyncSession
|
||||
|
||||
|
||||
def connector_binding(
|
||||
*,
|
||||
db_session: AsyncSession,
|
||||
search_space_id: int,
|
||||
user_id: str,
|
||||
) -> dict[str, AsyncSession | int | str]:
|
||||
return {
|
||||
"db_session": db_session,
|
||||
"search_space_id": search_space_id,
|
||||
"user_id": user_id,
|
||||
}
|
||||
|
|
@ -1,5 +0,0 @@
|
|||
"""Supervisor → domain message shaping."""
|
||||
|
||||
from app.agents.multi_agent_chat.core.delegation.child_task import compose_child_task
|
||||
|
||||
__all__ = ["compose_child_task"]
|
||||
|
|
@ -1,22 +0,0 @@
|
|||
"""Fold orchestrator-selected context into the single user message sent to a domain agent."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
|
||||
def compose_child_task(task: str, *, curated_context: str | None = None) -> str:
|
||||
"""Build the domain-agent user message: optional curated KB/context + task.
|
||||
|
||||
When ``curated_context`` is set (from supervisor/KB wiring), it is prepended so the
|
||||
child sees only what orchestration chose — not the full parent transcript.
|
||||
"""
|
||||
task = task.strip()
|
||||
if not curated_context or not curated_context.strip():
|
||||
return f"<delegated_task>\n{task}\n</delegated_task>"
|
||||
return (
|
||||
"<delegated_context>\n"
|
||||
f"{curated_context.strip()}\n"
|
||||
"</delegated_context>\n\n"
|
||||
"<delegated_task>\n"
|
||||
f"{task}\n"
|
||||
"</delegated_task>"
|
||||
)
|
||||
|
|
@ -1,5 +0,0 @@
|
|||
"""Parsing LangGraph invoke results."""
|
||||
|
||||
from app.agents.multi_agent_chat.core.invocation.output import extract_last_assistant_text
|
||||
|
||||
__all__ = ["extract_last_assistant_text"]
|
||||
|
|
@ -1,17 +0,0 @@
|
|||
"""Extract displayable text from a LangGraph agent ``invoke`` / ``ainvoke`` result."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any
|
||||
|
||||
|
||||
def extract_last_assistant_text(result: dict[str, Any]) -> str:
|
||||
"""Return the last message's string content, or ``\"\"`` if missing."""
|
||||
messages = result.get("messages") or []
|
||||
if not messages:
|
||||
return ""
|
||||
last = messages[-1]
|
||||
content = getattr(last, "content", None)
|
||||
if isinstance(content, str):
|
||||
return content
|
||||
return str(last)
|
||||
|
|
@ -1,135 +0,0 @@
|
|||
"""Partition MCP tools onto multi-agent expert routes (read-only; does not change the MCP loader).
|
||||
|
||||
Uses the same connector discovery shape as ``load_mcp_tools`` (copied query below). Tools come from
|
||||
``app.agents.new_chat.tools.mcp_tool.load_mcp_tools``; routing uses metadata already set there:
|
||||
|
||||
- HTTP tools: ``metadata["mcp_connector_id"]`` → DB connector row → expert route.
|
||||
- stdio tools: no connector id on the tool; ``metadata["mcp_connector_name"]`` → connector name map
|
||||
(duplicate names: last row wins — rare).
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
import logging
|
||||
from collections import defaultdict
|
||||
from collections.abc import Sequence
|
||||
from typing import Any
|
||||
|
||||
from langchain_core.tools import BaseTool
|
||||
from sqlalchemy import cast, select
|
||||
from sqlalchemy.dialects.postgresql import JSONB
|
||||
from sqlalchemy.ext.asyncio import AsyncSession
|
||||
|
||||
from app.db import SearchSourceConnector
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
# SurfSense ``SearchSourceConnectorType`` string → supervisor routing key (must match
|
||||
# ``DomainRoutingSpec.tool_name`` values used in ``supervisor_routing``).
|
||||
_CONNECTOR_TYPE_TO_EXPERT_ROUTE: dict[str, str] = {
|
||||
"GOOGLE_GMAIL_CONNECTOR": "gmail",
|
||||
"COMPOSIO_GMAIL_CONNECTOR": "gmail",
|
||||
"GOOGLE_CALENDAR_CONNECTOR": "calendar",
|
||||
"COMPOSIO_GOOGLE_CALENDAR_CONNECTOR": "calendar",
|
||||
"DISCORD_CONNECTOR": "discord",
|
||||
"TEAMS_CONNECTOR": "teams",
|
||||
"LUMA_CONNECTOR": "luma",
|
||||
"LINEAR_CONNECTOR": "linear",
|
||||
"JIRA_CONNECTOR": "jira",
|
||||
"CLICKUP_CONNECTOR": "clickup",
|
||||
"SLACK_CONNECTOR": "slack",
|
||||
"AIRTABLE_CONNECTOR": "airtable",
|
||||
# generic_mcp route intentionally disabled for now.
|
||||
# "MCP_CONNECTOR": "generic_mcp",
|
||||
}
|
||||
|
||||
# Ordering when appending MCP-only routes (no native registry slice for these types).
|
||||
MCP_ONLY_ROUTE_KEYS_IN_ORDER: tuple[str, ...] = (
|
||||
"linear",
|
||||
"slack",
|
||||
"jira",
|
||||
"clickup",
|
||||
"airtable",
|
||||
# generic_mcp intentionally disabled for now.
|
||||
# "generic_mcp",
|
||||
)
|
||||
|
||||
|
||||
async def fetch_mcp_connector_metadata_maps(
|
||||
session: AsyncSession,
|
||||
search_space_id: int,
|
||||
) -> tuple[dict[int, str], dict[str, str]]:
|
||||
"""Read-only copy of connector discovery used alongside ``load_mcp_tools``.
|
||||
|
||||
Same filter as :func:`app.agents.new_chat.tools.mcp_tool.load_mcp_tools` (connectors with ``server_config``).
|
||||
"""
|
||||
result = await session.execute(
|
||||
select(SearchSourceConnector).filter(
|
||||
SearchSourceConnector.search_space_id == search_space_id,
|
||||
cast(SearchSourceConnector.config, JSONB).has_key("server_config"),
|
||||
),
|
||||
)
|
||||
id_to_type: dict[int, str] = {}
|
||||
name_to_type: dict[str, str] = {}
|
||||
for connector in result.scalars():
|
||||
ct = (
|
||||
connector.connector_type.value
|
||||
if hasattr(connector.connector_type, "value")
|
||||
else str(connector.connector_type)
|
||||
)
|
||||
id_to_type[connector.id] = ct
|
||||
if connector.name:
|
||||
name_to_type[connector.name] = ct
|
||||
return id_to_type, name_to_type
|
||||
|
||||
|
||||
def partition_mcp_tools_by_expert_route(
|
||||
tools: Sequence[BaseTool],
|
||||
connector_id_to_type: dict[int, str],
|
||||
connector_name_to_type: dict[str, str],
|
||||
) -> dict[str, list[BaseTool]]:
|
||||
"""Bucket MCP tools by expert route key. Supervisor never receives raw MCP tools.
|
||||
|
||||
Same inclusion rule as :func:`app.agents.new_chat.tools.registry.build_tools_async`: all tools returned by
|
||||
``load_mcp_tools`` are partitioned — connector availability for **registry** builtins is handled via
|
||||
``get_connector_gated_tools`` / routing gates; MCP tools are not pre-filtered by inventory here.
|
||||
"""
|
||||
buckets: dict[str, list[BaseTool]] = defaultdict(list)
|
||||
|
||||
for tool in tools:
|
||||
meta: dict[str, Any] = getattr(tool, "metadata", None) or {}
|
||||
connector_type: str | None = None
|
||||
|
||||
cid = meta.get("mcp_connector_id")
|
||||
if cid is not None:
|
||||
try:
|
||||
cid_int = int(cid)
|
||||
except (TypeError, ValueError):
|
||||
cid_int = None
|
||||
if cid_int is not None:
|
||||
connector_type = connector_id_to_type.get(cid_int)
|
||||
|
||||
if connector_type is None and meta.get("mcp_transport") == "stdio":
|
||||
cname = meta.get("mcp_connector_name")
|
||||
if cname:
|
||||
connector_type = connector_name_to_type.get(str(cname))
|
||||
|
||||
if connector_type is None:
|
||||
logger.debug(
|
||||
"Skipping MCP tool %r — could not resolve connector type from metadata",
|
||||
getattr(tool, "name", None),
|
||||
)
|
||||
continue
|
||||
|
||||
route = _CONNECTOR_TYPE_TO_EXPERT_ROUTE.get(connector_type)
|
||||
if route is None:
|
||||
logger.warning(
|
||||
"MCP tool %r has unmapped connector type %s — skipped",
|
||||
getattr(tool, "name", None),
|
||||
connector_type,
|
||||
)
|
||||
continue
|
||||
|
||||
buckets[route].append(tool)
|
||||
|
||||
return dict(buckets)
|
||||
|
|
@ -1,5 +0,0 @@
|
|||
"""Markdown prompt loading for domain and supervisor packages."""
|
||||
|
||||
from app.agents.multi_agent_chat.core.prompts.load import read_prompt_md
|
||||
|
||||
__all__ = ["read_prompt_md"]
|
||||
|
|
@ -1,19 +0,0 @@
|
|||
"""Load ``*.md`` prompt files from co-located packages (domain slices ship ``domain_prompt.md``)."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from importlib import resources
|
||||
|
||||
|
||||
def read_prompt_md(package: str, stem: str) -> str:
|
||||
"""Read ``{stem}.md`` from the given import package (e.g. ``…expert_agent.connectors.notion``)."""
|
||||
try:
|
||||
ref = resources.files(package).joinpath(f"{stem}.md")
|
||||
if not ref.is_file():
|
||||
return ""
|
||||
text = ref.read_text(encoding="utf-8")
|
||||
except (FileNotFoundError, ModuleNotFoundError, OSError, TypeError):
|
||||
return ""
|
||||
if text.endswith("\n"):
|
||||
text = text[:-1]
|
||||
return text
|
||||
|
|
@ -1,15 +0,0 @@
|
|||
"""Main chat tool registry grouping + dependency bundles for domain slices."""
|
||||
|
||||
from app.agents.multi_agent_chat.core.registry.categories import (
|
||||
REGISTRY_ROUTING_CATEGORY_KEYS,
|
||||
TOOL_NAMES_BY_CATEGORY,
|
||||
)
|
||||
from app.agents.multi_agent_chat.core.registry.dependencies import build_registry_dependencies
|
||||
from app.agents.multi_agent_chat.core.registry.subset import build_registry_tools_for_category
|
||||
|
||||
__all__ = [
|
||||
"REGISTRY_ROUTING_CATEGORY_KEYS",
|
||||
"TOOL_NAMES_BY_CATEGORY",
|
||||
"build_registry_dependencies",
|
||||
"build_registry_tools_for_category",
|
||||
]
|
||||
|
|
@ -1,84 +0,0 @@
|
|||
"""Registry tool names grouped by multi-agent routing category.
|
||||
|
||||
Each string must match ``ToolDefinition.name`` in
|
||||
``app.agents.new_chat.tools.registry.BUILTIN_TOOLS`` — these are **not** guessed or MCP-only:
|
||||
:class:`~app.agents.multi_agent_chat.core.registry.subset.build_registry_tools_for_category`
|
||||
uses synchronous :func:`~app.agents.new_chat.tools.registry.build_tools`, which only instantiates
|
||||
``BUILTIN_TOOLS``. MCP tools are loaded separately and merged in ``supervisor_routing``.
|
||||
|
||||
Connectors that exist for search/indexing but have **no** entry in ``BUILTIN_TOOLS`` correctly have
|
||||
no row here (no chat tools to delegate)."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
# Keys match supervisor routing tool names; values match ``BUILTIN_TOOLS`` names exactly.
|
||||
TOOL_NAMES_BY_CATEGORY: dict[str, list[str]] = {
|
||||
"gmail": [
|
||||
"search_gmail",
|
||||
"read_gmail_email",
|
||||
"create_gmail_draft",
|
||||
"send_gmail_email",
|
||||
"trash_gmail_email",
|
||||
"update_gmail_draft",
|
||||
],
|
||||
"calendar": [
|
||||
"search_calendar_events",
|
||||
"create_calendar_event",
|
||||
"update_calendar_event",
|
||||
"delete_calendar_event",
|
||||
],
|
||||
"research": [
|
||||
"web_search",
|
||||
"scrape_webpage",
|
||||
"search_surfsense_docs",
|
||||
],
|
||||
"deliverables": [
|
||||
"generate_podcast",
|
||||
"generate_video_presentation",
|
||||
"generate_report",
|
||||
"generate_resume",
|
||||
"generate_image",
|
||||
],
|
||||
"memory": [
|
||||
"update_memory",
|
||||
],
|
||||
"discord": [
|
||||
"list_discord_channels",
|
||||
"read_discord_messages",
|
||||
"send_discord_message",
|
||||
],
|
||||
"teams": [
|
||||
"list_teams_channels",
|
||||
"read_teams_messages",
|
||||
"send_teams_message",
|
||||
],
|
||||
"notion": [
|
||||
"create_notion_page",
|
||||
"update_notion_page",
|
||||
"delete_notion_page",
|
||||
],
|
||||
"confluence": [
|
||||
"create_confluence_page",
|
||||
"update_confluence_page",
|
||||
"delete_confluence_page",
|
||||
],
|
||||
"google_drive": [
|
||||
"create_google_drive_file",
|
||||
"delete_google_drive_file",
|
||||
],
|
||||
"dropbox": [
|
||||
"create_dropbox_file",
|
||||
"delete_dropbox_file",
|
||||
],
|
||||
"onedrive": [
|
||||
"create_onedrive_file",
|
||||
"delete_onedrive_file",
|
||||
],
|
||||
"luma": [
|
||||
"list_luma_events",
|
||||
"read_luma_event",
|
||||
"create_luma_event",
|
||||
],
|
||||
}
|
||||
|
||||
REGISTRY_ROUTING_CATEGORY_KEYS: tuple[str, ...] = tuple(TOOL_NAMES_BY_CATEGORY.keys())
|
||||
|
|
@ -1,61 +0,0 @@
|
|||
"""Dependency dict for :func:`app.agents.new_chat.tools.registry.build_tools` on expert subgraphs."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any
|
||||
|
||||
from langchain_core.language_models import BaseChatModel
|
||||
from sqlalchemy.ext.asyncio import AsyncSession
|
||||
|
||||
from app.db import ChatVisibility
|
||||
|
||||
|
||||
def coerce_thread_id_for_registry(thread_id: str | int | None) -> int | None:
|
||||
"""Normalize chat thread id for registry tools that FK to ``new_chat_threads.id``.
|
||||
|
||||
``create_surfsense_deep_agent`` passes an ``int``; multi-agent wiring may pass
|
||||
``str(chat_id)`` for LangGraph/checkpointer consistency. AsyncPG requires ``int``
|
||||
for integer columns.
|
||||
"""
|
||||
if thread_id is None:
|
||||
return None
|
||||
if isinstance(thread_id, int):
|
||||
return thread_id
|
||||
s = str(thread_id).strip()
|
||||
if not s:
|
||||
return None
|
||||
if s.isdigit():
|
||||
return int(s)
|
||||
return None
|
||||
|
||||
|
||||
def build_registry_dependencies(
|
||||
*,
|
||||
db_session: AsyncSession,
|
||||
search_space_id: int,
|
||||
user_id: str,
|
||||
thread_id: str | int | None,
|
||||
llm: BaseChatModel | None = None,
|
||||
firecrawl_api_key: str | None = None,
|
||||
connector_service: Any | None = None,
|
||||
available_connectors: list[str] | None = None,
|
||||
available_document_types: list[str] | None = None,
|
||||
thread_visibility: ChatVisibility = ChatVisibility.PRIVATE,
|
||||
) -> dict[str, Any]:
|
||||
"""Union of kwargs commonly required by registry factories across category slices.
|
||||
|
||||
Individual categories enable a subset of tools; each tool still validates its own
|
||||
``ToolDefinition.requires`` against this dict.
|
||||
"""
|
||||
return {
|
||||
"db_session": db_session,
|
||||
"search_space_id": search_space_id,
|
||||
"user_id": user_id,
|
||||
"thread_id": coerce_thread_id_for_registry(thread_id),
|
||||
"llm": llm,
|
||||
"firecrawl_api_key": firecrawl_api_key,
|
||||
"connector_service": connector_service,
|
||||
"available_connectors": available_connectors,
|
||||
"available_document_types": available_document_types,
|
||||
"thread_visibility": thread_visibility,
|
||||
}
|
||||
|
|
@ -1,22 +0,0 @@
|
|||
"""Build registry tool subsets (``app.agents.new_chat.tools.registry``) for multi-agent domain slices."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any
|
||||
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.new_chat.tools.registry import build_tools
|
||||
from app.agents.multi_agent_chat.core.registry.categories import TOOL_NAMES_BY_CATEGORY
|
||||
|
||||
|
||||
def build_registry_tools_for_category(
|
||||
dependencies: dict[str, Any],
|
||||
category: str,
|
||||
) -> list[BaseTool]:
|
||||
"""Instantiate only the tools registered for ``category`` (see ``TOOL_NAMES_BY_CATEGORY``)."""
|
||||
names = TOOL_NAMES_BY_CATEGORY.get(category)
|
||||
if not names:
|
||||
msg = f"Unknown registry category: {category!r}"
|
||||
raise ValueError(msg)
|
||||
return build_tools(dependencies, enabled_tools=names)
|
||||
|
|
@ -1,5 +0,0 @@
|
|||
"""Expert subgraphs (specialists the supervisor delegates to).
|
||||
|
||||
- :mod:`expert_agent.builtins` — cross-cutting registry categories (e.g. research, memory, deliverables).
|
||||
- :mod:`expert_agent.connectors` — vendor/product integrations (email, chat, documents, … — one slice per route).
|
||||
"""
|
||||
|
|
@ -1 +0,0 @@
|
|||
"""Built-ins: broad capability categories from the registry (not single-vendor integrations)."""
|
||||
|
|
@ -1,11 +0,0 @@
|
|||
"""Deliverables vertical slice: registry tools, domain agent, ``domain_prompt.md``."""
|
||||
|
||||
from app.agents.multi_agent_chat.expert_agent.builtins.deliverables.agent import build_deliverables_domain_agent
|
||||
from app.agents.multi_agent_chat.expert_agent.builtins.deliverables.slice_tools import (
|
||||
build_deliverables_tools,
|
||||
)
|
||||
|
||||
__all__ = [
|
||||
"build_deliverables_tools",
|
||||
"build_deliverables_domain_agent",
|
||||
]
|
||||
|
|
@ -1,21 +0,0 @@
|
|||
"""Deliverables domain agent graph."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from collections.abc import Sequence
|
||||
|
||||
import app.agents.multi_agent_chat.expert_agent.builtins.deliverables as deliverables_pkg
|
||||
from langchain_core.language_models import BaseChatModel
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.agents import build_domain_agent
|
||||
|
||||
|
||||
def build_deliverables_domain_agent(llm: BaseChatModel, tools: Sequence[BaseTool]):
|
||||
"""Compiled deliverables domain-agent graph."""
|
||||
return build_domain_agent(
|
||||
llm,
|
||||
tools,
|
||||
prompt_package=deliverables_pkg.__name__,
|
||||
prompt_stem="domain_prompt",
|
||||
)
|
||||
|
|
@ -1,55 +0,0 @@
|
|||
You are the SurfSense deliverables operations sub-agent.
|
||||
You receive delegated instructions from a supervisor agent and return structured results for supervisor synthesis.
|
||||
|
||||
<goal>
|
||||
Produce **deliverables**: shareable **artifacts** the user keeps (reports, slide-style video presentations, podcasts, resumes, images). Use explicit constraints and reliable proof of what was generated.
|
||||
</goal>
|
||||
|
||||
<available_tools>
|
||||
- `generate_report`
|
||||
- `generate_podcast`
|
||||
- `generate_video_presentation`
|
||||
- `generate_resume`
|
||||
- `generate_image`
|
||||
</available_tools>
|
||||
|
||||
<tool_policy>
|
||||
- Use only tools in `<available_tools>`.
|
||||
- Require essential generation constraints (audience, format, tone, core content).
|
||||
- If critical constraints are missing, return `status=blocked` with `missing_fields`.
|
||||
- Never claim artifact generation success without tool confirmation.
|
||||
</tool_policy>
|
||||
|
||||
<out_of_scope>
|
||||
- Do not perform connector data mutations unrelated to artifact generation.
|
||||
</out_of_scope>
|
||||
|
||||
<safety>
|
||||
- Avoid generating artifacts with missing critical constraints.
|
||||
- Prefer one complete artifact over partial multi-artifact output.
|
||||
</safety>
|
||||
|
||||
<failure_policy>
|
||||
- On generation failure, return `status=error` with best retry guidance.
|
||||
- On missing constraints, return `status=blocked` with required fields.
|
||||
</failure_policy>
|
||||
|
||||
<output_contract>
|
||||
Return **only** one JSON object (no markdown/prose):
|
||||
{
|
||||
"status": "success" | "partial" | "blocked" | "error",
|
||||
"action_summary": string,
|
||||
"evidence": {
|
||||
"artifact_type": "report" | "podcast" | "video_presentation" | "resume" | "image" | null,
|
||||
"artifact_id": string | null,
|
||||
"artifact_location": string | null
|
||||
},
|
||||
"next_step": string | null,
|
||||
"missing_fields": string[] | null,
|
||||
"assumptions": string[] | null
|
||||
}
|
||||
Rules:
|
||||
- `status=success` -> `next_step=null`, `missing_fields=null`.
|
||||
- `status=partial|blocked|error` -> `next_step` must be non-null.
|
||||
- `status=blocked` due to missing required inputs -> `missing_fields` must be non-null.
|
||||
</output_contract>
|
||||
|
|
@ -1,14 +0,0 @@
|
|||
"""Registry-backed deliverables tools (reports, media exports, resume, images)."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any
|
||||
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.registry import build_registry_tools_for_category
|
||||
|
||||
|
||||
def build_deliverables_tools(dependencies: dict[str, Any]) -> list[BaseTool]:
|
||||
"""Registry-backed tools for the ``deliverables`` category."""
|
||||
return build_registry_tools_for_category(dependencies, "deliverables")
|
||||
|
|
@ -1,9 +0,0 @@
|
|||
"""Memory vertical slice: registry tools, domain agent, ``domain_prompt.md``."""
|
||||
|
||||
from app.agents.multi_agent_chat.expert_agent.builtins.memory.agent import build_memory_domain_agent
|
||||
from app.agents.multi_agent_chat.expert_agent.builtins.memory.slice_tools import build_memory_tools
|
||||
|
||||
__all__ = [
|
||||
"build_memory_tools",
|
||||
"build_memory_domain_agent",
|
||||
]
|
||||
|
|
@ -1,45 +0,0 @@
|
|||
"""Memory domain agent graph."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from collections.abc import Sequence
|
||||
|
||||
import app.agents.multi_agent_chat.expert_agent.builtins.memory as memory_pkg
|
||||
from langchain.agents import create_agent
|
||||
from langchain_core.language_models import BaseChatModel
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.prompts import read_prompt_md
|
||||
from app.db import ChatVisibility
|
||||
|
||||
_PRIVATE_VISIBILITY_POLICY = (
|
||||
"This thread is private. Store user-specific long-lived preferences, facts, and instructions."
|
||||
)
|
||||
_TEAM_VISIBILITY_POLICY = (
|
||||
"This thread is shared with the search space. Store only team-appropriate shared preferences,"
|
||||
" facts, and instructions that are safe for all members to inherit."
|
||||
)
|
||||
|
||||
|
||||
def _render_memory_prompt(thread_visibility: ChatVisibility | None) -> str:
|
||||
template = read_prompt_md(memory_pkg.__name__, "domain_prompt")
|
||||
policy = (
|
||||
_TEAM_VISIBILITY_POLICY
|
||||
if thread_visibility == ChatVisibility.SEARCH_SPACE
|
||||
else _PRIVATE_VISIBILITY_POLICY
|
||||
)
|
||||
return template.replace("{{MEMORY_VISIBILITY_POLICY}}", policy)
|
||||
|
||||
|
||||
def build_memory_domain_agent(
|
||||
llm: BaseChatModel,
|
||||
tools: Sequence[BaseTool],
|
||||
*,
|
||||
thread_visibility: ChatVisibility | None = None,
|
||||
):
|
||||
"""Compiled memory domain-agent graph."""
|
||||
return create_agent(
|
||||
llm,
|
||||
system_prompt=_render_memory_prompt(thread_visibility),
|
||||
tools=list(tools),
|
||||
)
|
||||
|
|
@ -1,56 +0,0 @@
|
|||
You are the SurfSense memory operations sub-agent.
|
||||
You receive delegated instructions from a supervisor agent and return structured results for supervisor synthesis.
|
||||
|
||||
<goal>
|
||||
Persist durable preferences/facts/instructions with `update_memory` while avoiding transient or unsafe storage.
|
||||
</goal>
|
||||
|
||||
<visibility_scope>
|
||||
{{MEMORY_VISIBILITY_POLICY}}
|
||||
</visibility_scope>
|
||||
|
||||
<available_tools>
|
||||
- `update_memory`
|
||||
</available_tools>
|
||||
|
||||
<tool_policy>
|
||||
- Save only durable information with future value.
|
||||
- Do not store transient chatter.
|
||||
- Do not store secrets unless explicitly instructed.
|
||||
- If memory intent is unclear, return `status=blocked` with the missing intent signal.
|
||||
</tool_policy>
|
||||
|
||||
<out_of_scope>
|
||||
- Do not execute non-memory tool actions.
|
||||
- Do not store irrelevant, transient, or speculative information.
|
||||
</out_of_scope>
|
||||
|
||||
<safety>
|
||||
- Prefer minimal-memory writes over over-collection.
|
||||
- Never claim memory was updated unless `update_memory` succeeded.
|
||||
</safety>
|
||||
|
||||
<failure_policy>
|
||||
- On tool failure, return `status=error` with concise recovery steps.
|
||||
- When intent is ambiguous, return `status=blocked` with required disambiguation fields.
|
||||
</failure_policy>
|
||||
|
||||
<output_contract>
|
||||
Return **only** one JSON object (no markdown/prose):
|
||||
{
|
||||
"status": "success" | "partial" | "blocked" | "error",
|
||||
"action_summary": string,
|
||||
"evidence": {
|
||||
"memory_updated": boolean,
|
||||
"memory_category": "preference" | "fact" | "instruction" | null,
|
||||
"stored_summary": string | null
|
||||
},
|
||||
"next_step": string | null,
|
||||
"missing_fields": string[] | null,
|
||||
"assumptions": string[] | null
|
||||
}
|
||||
Rules:
|
||||
- `status=success` -> `next_step=null`, `missing_fields=null`.
|
||||
- `status=partial|blocked|error` -> `next_step` must be non-null.
|
||||
- `status=blocked` due to missing required inputs -> `missing_fields` must be non-null.
|
||||
</output_contract>
|
||||
|
|
@ -1,14 +0,0 @@
|
|||
"""Registry-backed memory tools (long-term user or team memory)."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any
|
||||
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.registry import build_registry_tools_for_category
|
||||
|
||||
|
||||
def build_memory_tools(dependencies: dict[str, Any]) -> list[BaseTool]:
|
||||
"""Registry-backed tools for the ``memory`` category."""
|
||||
return build_registry_tools_for_category(dependencies, "memory")
|
||||
|
|
@ -1,9 +0,0 @@
|
|||
"""Research vertical slice: registry tools, domain agent, ``domain_prompt.md``."""
|
||||
|
||||
from app.agents.multi_agent_chat.expert_agent.builtins.research.agent import build_research_domain_agent
|
||||
from app.agents.multi_agent_chat.expert_agent.builtins.research.slice_tools import build_research_tools
|
||||
|
||||
__all__ = [
|
||||
"build_research_tools",
|
||||
"build_research_domain_agent",
|
||||
]
|
||||
|
|
@ -1,21 +0,0 @@
|
|||
"""Research domain agent graph."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from collections.abc import Sequence
|
||||
|
||||
import app.agents.multi_agent_chat.expert_agent.builtins.research as research_pkg
|
||||
from langchain_core.language_models import BaseChatModel
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.agents import build_domain_agent
|
||||
|
||||
|
||||
def build_research_domain_agent(llm: BaseChatModel, tools: Sequence[BaseTool]):
|
||||
"""Compiled research domain-agent graph."""
|
||||
return build_domain_agent(
|
||||
llm,
|
||||
tools,
|
||||
prompt_package=research_pkg.__name__,
|
||||
prompt_stem="domain_prompt",
|
||||
)
|
||||
|
|
@ -1,53 +0,0 @@
|
|||
You are the SurfSense research operations sub-agent.
|
||||
You receive delegated instructions from a supervisor agent and return structured results for supervisor synthesis.
|
||||
|
||||
<goal>
|
||||
Gather and synthesize evidence using SurfSense research tools with clear citations and uncertainty reporting.
|
||||
</goal>
|
||||
|
||||
<available_tools>
|
||||
- `web_search`
|
||||
- `scrape_webpage`
|
||||
- `search_surfsense_docs`
|
||||
</available_tools>
|
||||
|
||||
<tool_policy>
|
||||
- Use only tools in `<available_tools>`.
|
||||
- Prefer primary and recent sources when recency matters.
|
||||
- If the delegated request is underspecified, return `status=blocked` with the missing research constraints.
|
||||
- Never fabricate facts, citations, URLs, or quote text.
|
||||
</tool_policy>
|
||||
|
||||
<out_of_scope>
|
||||
- Do not execute connector mutations (email/calendar/docs/chat writes) or deliverable generation.
|
||||
</out_of_scope>
|
||||
|
||||
<safety>
|
||||
- Report uncertainty explicitly when evidence is incomplete or conflicting.
|
||||
- Never present unverified claims as facts.
|
||||
</safety>
|
||||
|
||||
<failure_policy>
|
||||
- On tool failure, return `status=error` with a concise recovery `next_step`.
|
||||
- On no useful evidence, return `status=blocked` with recommended narrower filters.
|
||||
</failure_policy>
|
||||
|
||||
<output_contract>
|
||||
Return **only** one JSON object (no markdown/prose):
|
||||
{
|
||||
"status": "success" | "partial" | "blocked" | "error",
|
||||
"action_summary": string,
|
||||
"evidence": {
|
||||
"findings": string[],
|
||||
"sources": string[],
|
||||
"confidence": "high" | "medium" | "low"
|
||||
},
|
||||
"next_step": string | null,
|
||||
"missing_fields": string[] | null,
|
||||
"assumptions": string[] | null
|
||||
}
|
||||
Rules:
|
||||
- `status=success` -> `next_step=null`, `missing_fields=null`.
|
||||
- `status=partial|blocked|error` -> `next_step` must be non-null.
|
||||
- `status=blocked` due to missing required inputs -> `missing_fields` must be non-null.
|
||||
</output_contract>
|
||||
|
|
@ -1,14 +0,0 @@
|
|||
"""Registry-backed research tools (web, scrape, SurfSense docs help)."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any
|
||||
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.registry import build_registry_tools_for_category
|
||||
|
||||
|
||||
def build_research_tools(dependencies: dict[str, Any]) -> list[BaseTool]:
|
||||
"""Registry-backed tools for the ``research`` category."""
|
||||
return build_registry_tools_for_category(dependencies, "research")
|
||||
|
|
@ -1 +0,0 @@
|
|||
"""External integrations: third-party products (explicit factories or registry-backed connector tools)."""
|
||||
|
|
@ -1,11 +0,0 @@
|
|||
"""Google Calendar vertical slice: registry tools, domain agent, ``domain_prompt.md``."""
|
||||
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.calendar.agent import build_calendar_domain_agent
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.calendar.slice_tools import (
|
||||
build_calendar_tools,
|
||||
)
|
||||
|
||||
__all__ = [
|
||||
"build_calendar_domain_agent",
|
||||
"build_calendar_tools",
|
||||
]
|
||||
|
|
@ -1,21 +0,0 @@
|
|||
"""Google Calendar domain agent graph."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from collections.abc import Sequence
|
||||
|
||||
import app.agents.multi_agent_chat.expert_agent.connectors.calendar as calendar_pkg
|
||||
from langchain_core.language_models import BaseChatModel
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.agents import build_domain_agent
|
||||
|
||||
|
||||
def build_calendar_domain_agent(llm: BaseChatModel, tools: Sequence[BaseTool]):
|
||||
"""Compiled Google Calendar domain-agent graph."""
|
||||
return build_domain_agent(
|
||||
llm,
|
||||
tools,
|
||||
prompt_package=calendar_pkg.__name__,
|
||||
prompt_stem="domain_prompt",
|
||||
)
|
||||
|
|
@ -1,62 +0,0 @@
|
|||
You are the Google Calendar operations sub-agent.
|
||||
You receive delegated instructions from a supervisor agent and return structured results for supervisor synthesis.
|
||||
|
||||
<goal>
|
||||
Execute calendar event operations (search, create, update, delete) accurately with timezone-safe scheduling.
|
||||
</goal>
|
||||
|
||||
<available_tools>
|
||||
- `search_calendar_events`
|
||||
- `create_calendar_event`
|
||||
- `update_calendar_event`
|
||||
- `delete_calendar_event`
|
||||
</available_tools>
|
||||
|
||||
<tool_policy>
|
||||
- Use only tools in `<available_tools>`.
|
||||
- Resolve relative dates against current runtime timestamp.
|
||||
- If required fields (date/time/timezone/target event) are missing or ambiguous, return `status=blocked` with `missing_fields` and supervisor `next_step`.
|
||||
- Never invent event IDs or mutation results.
|
||||
</tool_policy>
|
||||
|
||||
<out_of_scope>
|
||||
- Do not perform non-calendar tasks.
|
||||
</out_of_scope>
|
||||
|
||||
<safety>
|
||||
- Before update/delete, ensure event target is explicit.
|
||||
- Never claim event mutation success without tool confirmation.
|
||||
</safety>
|
||||
|
||||
<failure_policy>
|
||||
- On tool failure, return `status=error` with concise recovery `next_step`.
|
||||
- On ambiguity, return `status=blocked` with top event candidates.
|
||||
</failure_policy>
|
||||
|
||||
<output_contract>
|
||||
Return **only** one JSON object (no markdown/prose):
|
||||
{
|
||||
"status": "success" | "partial" | "blocked" | "error",
|
||||
"action_summary": string,
|
||||
"evidence": {
|
||||
"event_id": string | null,
|
||||
"title": string | null,
|
||||
"start_at": string (ISO 8601 with timezone) | null,
|
||||
"end_at": string (ISO 8601 with timezone) | null,
|
||||
"matched_candidates": [
|
||||
{
|
||||
"event_id": string,
|
||||
"title": string | null,
|
||||
"start_at": string (ISO 8601 with timezone) | null
|
||||
}
|
||||
] | null
|
||||
},
|
||||
"next_step": string | null,
|
||||
"missing_fields": string[] | null,
|
||||
"assumptions": string[] | null
|
||||
}
|
||||
Rules:
|
||||
- `status=success` -> `next_step=null`, `missing_fields=null`.
|
||||
- `status=partial|blocked|error` -> `next_step` must be non-null.
|
||||
- `status=blocked` due to missing required inputs -> `missing_fields` must be non-null.
|
||||
</output_contract>
|
||||
|
|
@ -1,14 +0,0 @@
|
|||
"""Registry-backed Google Calendar tools."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any
|
||||
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.registry import build_registry_tools_for_category
|
||||
|
||||
|
||||
def build_calendar_tools(dependencies: dict[str, Any]) -> list[BaseTool]:
|
||||
"""Registry-backed tools for the ``calendar`` category."""
|
||||
return build_registry_tools_for_category(dependencies, "calendar")
|
||||
|
|
@ -1,11 +0,0 @@
|
|||
"""Confluence connector slice."""
|
||||
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.confluence.agent import build_confluence_domain_agent
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.confluence.slice_tools import (
|
||||
build_confluence_tools,
|
||||
)
|
||||
|
||||
__all__ = [
|
||||
"build_confluence_tools",
|
||||
"build_confluence_domain_agent",
|
||||
]
|
||||
|
|
@ -1,21 +0,0 @@
|
|||
"""Confluence domain agent graph."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from collections.abc import Sequence
|
||||
|
||||
import app.agents.multi_agent_chat.expert_agent.connectors.confluence as confluence_pkg
|
||||
from langchain_core.language_models import BaseChatModel
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.agents import build_domain_agent
|
||||
|
||||
|
||||
def build_confluence_domain_agent(llm: BaseChatModel, tools: Sequence[BaseTool]):
|
||||
"""Compiled Confluence domain-agent graph."""
|
||||
return build_domain_agent(
|
||||
llm,
|
||||
tools,
|
||||
prompt_package=confluence_pkg.__name__,
|
||||
prompt_stem="domain_prompt",
|
||||
)
|
||||
|
|
@ -1,55 +0,0 @@
|
|||
You are the Confluence operations sub-agent.
|
||||
You receive delegated instructions from a supervisor agent and return structured results for supervisor synthesis.
|
||||
|
||||
<goal>
|
||||
Execute Confluence page operations accurately in the connected space.
|
||||
</goal>
|
||||
|
||||
<available_tools>
|
||||
- `create_confluence_page`
|
||||
- `update_confluence_page`
|
||||
- `delete_confluence_page`
|
||||
</available_tools>
|
||||
|
||||
<tool_policy>
|
||||
- Use only tools in `<available_tools>`.
|
||||
- Verify target page and intended mutation before update/delete.
|
||||
- If target page is ambiguous, return `status=blocked` with candidate options for supervisor disambiguation.
|
||||
- Never invent page IDs, titles, or mutation outcomes.
|
||||
</tool_policy>
|
||||
|
||||
<out_of_scope>
|
||||
- Do not perform non-Confluence tasks.
|
||||
</out_of_scope>
|
||||
|
||||
<safety>
|
||||
- Never claim page mutation success without tool confirmation.
|
||||
- If destructive action appears already completed in this session, do not repeat; return prior evidence with an `assumptions` note.
|
||||
</safety>
|
||||
|
||||
<failure_policy>
|
||||
- On tool failure, return `status=error` with concise retry/recovery `next_step`.
|
||||
- On unresolved page ambiguity, return `status=blocked` with candidates.
|
||||
</failure_policy>
|
||||
|
||||
<output_contract>
|
||||
Return **only** one JSON object (no markdown/prose):
|
||||
{
|
||||
"status": "success" | "partial" | "blocked" | "error",
|
||||
"action_summary": string,
|
||||
"evidence": {
|
||||
"page_id": string | null,
|
||||
"page_title": string | null,
|
||||
"matched_candidates": [
|
||||
{ "page_id": string, "page_title": string | null }
|
||||
] | null
|
||||
},
|
||||
"next_step": string | null,
|
||||
"missing_fields": string[] | null,
|
||||
"assumptions": string[] | null
|
||||
}
|
||||
Rules:
|
||||
- `status=success` -> `next_step=null`, `missing_fields=null`.
|
||||
- `status=partial|blocked|error` -> `next_step` must be non-null.
|
||||
- `status=blocked` due to missing required inputs -> `missing_fields` must be non-null.
|
||||
</output_contract>
|
||||
|
|
@ -1,14 +0,0 @@
|
|||
"""Registry-backed Confluence tools."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any
|
||||
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.registry import build_registry_tools_for_category
|
||||
|
||||
|
||||
def build_confluence_tools(dependencies: dict[str, Any]) -> list[BaseTool]:
|
||||
"""Registry-backed tools for the ``confluence`` category."""
|
||||
return build_registry_tools_for_category(dependencies, "confluence")
|
||||
|
|
@ -1,9 +0,0 @@
|
|||
"""Discord vertical slice: registry tools, domain agent, ``domain_prompt.md``."""
|
||||
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.discord.agent import build_discord_domain_agent
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.discord.slice_tools import build_discord_tools
|
||||
|
||||
__all__ = [
|
||||
"build_discord_tools",
|
||||
"build_discord_domain_agent",
|
||||
]
|
||||
|
|
@ -1,21 +0,0 @@
|
|||
"""Discord domain agent graph."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from collections.abc import Sequence
|
||||
|
||||
import app.agents.multi_agent_chat.expert_agent.connectors.discord as discord_pkg
|
||||
from langchain_core.language_models import BaseChatModel
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.agents import build_domain_agent
|
||||
|
||||
|
||||
def build_discord_domain_agent(llm: BaseChatModel, tools: Sequence[BaseTool]):
|
||||
"""Compiled Discord domain-agent graph."""
|
||||
return build_domain_agent(
|
||||
llm,
|
||||
tools,
|
||||
prompt_package=discord_pkg.__name__,
|
||||
prompt_stem="domain_prompt",
|
||||
)
|
||||
|
|
@ -1,56 +0,0 @@
|
|||
You are the Discord operations sub-agent.
|
||||
You receive delegated instructions from a supervisor agent and return structured results for supervisor synthesis.
|
||||
|
||||
<goal>
|
||||
Execute Discord reads and sends accurately in the connected server/workspace.
|
||||
</goal>
|
||||
|
||||
<available_tools>
|
||||
- `list_discord_channels`
|
||||
- `read_discord_messages`
|
||||
- `send_discord_message`
|
||||
</available_tools>
|
||||
|
||||
<tool_policy>
|
||||
- Use only tools in `<available_tools>`.
|
||||
- Resolve channel/thread targets before reads/sends.
|
||||
- If target is ambiguous, return `status=blocked` with candidate channels/threads.
|
||||
- Never invent message content, sender identity, timestamps, or delivery results.
|
||||
</tool_policy>
|
||||
|
||||
<out_of_scope>
|
||||
- Do not perform non-Discord tasks.
|
||||
</out_of_scope>
|
||||
|
||||
<safety>
|
||||
- Before send, verify destination and message intent match delegated instructions.
|
||||
- Never claim send success without tool confirmation.
|
||||
</safety>
|
||||
|
||||
<failure_policy>
|
||||
- On tool failure, return `status=error` with concise recovery `next_step`.
|
||||
- On unresolved destination ambiguity, return `status=blocked` with candidate options.
|
||||
</failure_policy>
|
||||
|
||||
<output_contract>
|
||||
Return **only** one JSON object (no markdown/prose):
|
||||
{
|
||||
"status": "success" | "partial" | "blocked" | "error",
|
||||
"action_summary": string,
|
||||
"evidence": {
|
||||
"channel_id": string | null,
|
||||
"thread_id": string | null,
|
||||
"message_id": string | null,
|
||||
"matched_candidates": [
|
||||
{ "channel_id": string, "thread_id": string | null, "label": string | null }
|
||||
] | null
|
||||
},
|
||||
"next_step": string | null,
|
||||
"missing_fields": string[] | null,
|
||||
"assumptions": string[] | null
|
||||
}
|
||||
Rules:
|
||||
- `status=success` -> `next_step=null`, `missing_fields=null`.
|
||||
- `status=partial|blocked|error` -> `next_step` must be non-null.
|
||||
- `status=blocked` due to missing required inputs -> `missing_fields` must be non-null.
|
||||
</output_contract>
|
||||
|
|
@ -1,14 +0,0 @@
|
|||
"""Registry-backed Discord tools."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any
|
||||
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.registry import build_registry_tools_for_category
|
||||
|
||||
|
||||
def build_discord_tools(dependencies: dict[str, Any]) -> list[BaseTool]:
|
||||
"""Registry-backed tools for the ``discord`` category."""
|
||||
return build_registry_tools_for_category(dependencies, "discord")
|
||||
|
|
@ -1,11 +0,0 @@
|
|||
"""Dropbox connector slice."""
|
||||
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.dropbox.agent import build_dropbox_domain_agent
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.dropbox.slice_tools import (
|
||||
build_dropbox_tools,
|
||||
)
|
||||
|
||||
__all__ = [
|
||||
"build_dropbox_tools",
|
||||
"build_dropbox_domain_agent",
|
||||
]
|
||||
|
|
@ -1,21 +0,0 @@
|
|||
"""Dropbox domain agent graph."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from collections.abc import Sequence
|
||||
|
||||
import app.agents.multi_agent_chat.expert_agent.connectors.dropbox as dropbox_pkg
|
||||
from langchain_core.language_models import BaseChatModel
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.agents import build_domain_agent
|
||||
|
||||
|
||||
def build_dropbox_domain_agent(llm: BaseChatModel, tools: Sequence[BaseTool]):
|
||||
"""Compiled Dropbox domain-agent graph."""
|
||||
return build_domain_agent(
|
||||
llm,
|
||||
tools,
|
||||
prompt_package=dropbox_pkg.__name__,
|
||||
prompt_stem="domain_prompt",
|
||||
)
|
||||
|
|
@ -1,52 +0,0 @@
|
|||
You are the Dropbox operations sub-agent.
|
||||
You receive delegated instructions from a supervisor agent and return structured results for supervisor synthesis.
|
||||
|
||||
<goal>
|
||||
Execute Dropbox file create/delete actions accurately in the connected account.
|
||||
</goal>
|
||||
|
||||
<available_tools>
|
||||
- `create_dropbox_file`
|
||||
- `delete_dropbox_file`
|
||||
</available_tools>
|
||||
|
||||
<tool_policy>
|
||||
- Use only tools in `<available_tools>`.
|
||||
- Ensure target path/file identity is explicit before mutate actions.
|
||||
- If target is ambiguous, return `status=blocked` with candidate paths.
|
||||
- Never invent file IDs/paths or mutation outcomes.
|
||||
</tool_policy>
|
||||
|
||||
<out_of_scope>
|
||||
- Do not perform non-Dropbox tasks.
|
||||
</out_of_scope>
|
||||
|
||||
<safety>
|
||||
- Never claim file mutation success without tool confirmation.
|
||||
</safety>
|
||||
|
||||
<failure_policy>
|
||||
- On tool failure, return `status=error` with concise recovery `next_step`.
|
||||
- On target ambiguity, return `status=blocked` with candidate paths.
|
||||
</failure_policy>
|
||||
|
||||
<output_contract>
|
||||
Return **only** one JSON object (no markdown/prose):
|
||||
{
|
||||
"status": "success" | "partial" | "blocked" | "error",
|
||||
"action_summary": string,
|
||||
"evidence": {
|
||||
"file_path": string | null,
|
||||
"file_id": string | null,
|
||||
"operation": "create" | "delete" | null,
|
||||
"matched_candidates": string[] | null
|
||||
},
|
||||
"next_step": string | null,
|
||||
"missing_fields": string[] | null,
|
||||
"assumptions": string[] | null
|
||||
}
|
||||
Rules:
|
||||
- `status=success` -> `next_step=null`, `missing_fields=null`.
|
||||
- `status=partial|blocked|error` -> `next_step` must be non-null.
|
||||
- `status=blocked` due to missing required inputs -> `missing_fields` must be non-null.
|
||||
</output_contract>
|
||||
|
|
@ -1,14 +0,0 @@
|
|||
"""Registry-backed Dropbox tools."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any
|
||||
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.registry import build_registry_tools_for_category
|
||||
|
||||
|
||||
def build_dropbox_tools(dependencies: dict[str, Any]) -> list[BaseTool]:
|
||||
"""Registry-backed tools for the ``dropbox`` category."""
|
||||
return build_registry_tools_for_category(dependencies, "dropbox")
|
||||
|
|
@ -1,9 +0,0 @@
|
|||
"""Gmail vertical slice: registry tools, domain agent, ``domain_prompt.md``."""
|
||||
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.gmail.agent import build_gmail_domain_agent
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.gmail.slice_tools import build_gmail_tools
|
||||
|
||||
__all__ = [
|
||||
"build_gmail_tools",
|
||||
"build_gmail_domain_agent",
|
||||
]
|
||||
|
|
@ -1,21 +0,0 @@
|
|||
"""Gmail domain agent graph."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from collections.abc import Sequence
|
||||
|
||||
import app.agents.multi_agent_chat.expert_agent.connectors.gmail as gmail_pkg
|
||||
from langchain_core.language_models import BaseChatModel
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.agents import build_domain_agent
|
||||
|
||||
|
||||
def build_gmail_domain_agent(llm: BaseChatModel, tools: Sequence[BaseTool]):
|
||||
"""Compiled Gmail domain-agent graph."""
|
||||
return build_domain_agent(
|
||||
llm,
|
||||
tools,
|
||||
prompt_package=gmail_pkg.__name__,
|
||||
prompt_stem="domain_prompt",
|
||||
)
|
||||
|
|
@ -1,82 +0,0 @@
|
|||
You are the Gmail operations sub-agent.
|
||||
You receive delegated instructions from a supervisor agent and return structured results for supervisor synthesis.
|
||||
|
||||
<goal>
|
||||
Execute Gmail operations accurately: search/read emails, prepare drafts, send, and trash.
|
||||
</goal>
|
||||
|
||||
<available_tools>
|
||||
- `search_gmail`: find candidate emails with query constraints.
|
||||
- `read_gmail_email`: read one message in full detail.
|
||||
- `create_gmail_draft`: create a new draft.
|
||||
- `update_gmail_draft`: modify an existing draft.
|
||||
- `send_gmail_email`: send an email.
|
||||
- `trash_gmail_email`: move an email to trash.
|
||||
</available_tools>
|
||||
|
||||
<tool_policy>
|
||||
- Use only tools in `<available_tools>`.
|
||||
- Build precise search queries using Gmail operators when possible (`from:`, `to:`, `subject:`, `after:`, `before:`, `has:attachment`, `is:unread`, `label:`).
|
||||
- Resolve relative dates against runtime timestamp; prefer narrower interpretation.
|
||||
- For reply requests, identify the target thread/email via search + read before drafting.
|
||||
- If required fields are missing or target selection is ambiguous, return `status=blocked` with `missing_fields` and disambiguation candidates.
|
||||
- Never invent IDs, recipients, timestamps, quoted text, or tool outcomes.
|
||||
</tool_policy>
|
||||
|
||||
<out_of_scope>
|
||||
- Do not perform non-Gmail work.
|
||||
- Filing operations not represented in `<available_tools>` (archive/label/mark-read/move-folder) are unsupported here.
|
||||
</out_of_scope>
|
||||
|
||||
<safety>
|
||||
- For send: verify draft `to`, `subject`, and `body` match delegated instructions.
|
||||
- If any send-critical field was inferred, do not send; return `status=blocked` with inferred values in `assumptions`.
|
||||
- For trash: ensure explicit target match before deletion.
|
||||
- If a destructive action appears already completed this session, do not repeat; return prior evidence.
|
||||
</safety>
|
||||
|
||||
<failure_policy>
|
||||
- On tool failure, return `status=error` with concise recovery `next_step`.
|
||||
- If search has no strong match, return `status=blocked` with suggested tighter filters.
|
||||
- If multiple strong candidates remain for risky actions, return `status=blocked` with top options.
|
||||
</failure_policy>
|
||||
|
||||
<output_contract>
|
||||
Return **only** one JSON object (no markdown/prose):
|
||||
{
|
||||
"status": "success" | "partial" | "blocked" | "error",
|
||||
"action_summary": string,
|
||||
"evidence": {
|
||||
"email_id": string | null,
|
||||
"thread_id": string | null,
|
||||
"subject": string | null,
|
||||
"sender": string | null,
|
||||
"recipients": string[] | null,
|
||||
"received_at": string (ISO 8601 with timezone) | null,
|
||||
"sent_message": {
|
||||
"id": string,
|
||||
"to": string[],
|
||||
"subject": string | null,
|
||||
"sent_at": string (ISO 8601 with timezone) | null
|
||||
} | null,
|
||||
"matched_candidates": [
|
||||
{
|
||||
"email_id": string,
|
||||
"subject": string | null,
|
||||
"sender": string | null,
|
||||
"received_at": string (ISO 8601 with timezone) | null
|
||||
}
|
||||
] | null
|
||||
},
|
||||
"next_step": string | null,
|
||||
"missing_fields": string[] | null,
|
||||
"assumptions": string[] | null
|
||||
}
|
||||
|
||||
Rules:
|
||||
- `status=success` -> `next_step=null`, `missing_fields=null`.
|
||||
- `status=partial|blocked|error` -> `next_step` must be non-null.
|
||||
- `status=blocked` due to missing required inputs -> `missing_fields` must be non-null.
|
||||
- For blocked ambiguity, include options in `evidence.matched_candidates`.
|
||||
- For trash actions, `evidence.email_id` is the trashed message.
|
||||
</output_contract>
|
||||
|
|
@ -1,14 +0,0 @@
|
|||
"""Registry-backed Gmail tools."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any
|
||||
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.registry import build_registry_tools_for_category
|
||||
|
||||
|
||||
def build_gmail_tools(dependencies: dict[str, Any]) -> list[BaseTool]:
|
||||
"""Registry-backed tools for the ``gmail`` category."""
|
||||
return build_registry_tools_for_category(dependencies, "gmail")
|
||||
|
|
@ -1,13 +0,0 @@
|
|||
"""Google Drive connector slice."""
|
||||
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.google_drive.agent import (
|
||||
build_google_drive_domain_agent,
|
||||
)
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.google_drive.slice_tools import (
|
||||
build_google_drive_tools,
|
||||
)
|
||||
|
||||
__all__ = [
|
||||
"build_google_drive_tools",
|
||||
"build_google_drive_domain_agent",
|
||||
]
|
||||
|
|
@ -1,21 +0,0 @@
|
|||
"""Google Drive domain agent graph."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from collections.abc import Sequence
|
||||
|
||||
import app.agents.multi_agent_chat.expert_agent.connectors.google_drive as google_drive_pkg
|
||||
from langchain_core.language_models import BaseChatModel
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.agents import build_domain_agent
|
||||
|
||||
|
||||
def build_google_drive_domain_agent(llm: BaseChatModel, tools: Sequence[BaseTool]):
|
||||
"""Compiled Google Drive domain-agent graph."""
|
||||
return build_domain_agent(
|
||||
llm,
|
||||
tools,
|
||||
prompt_package=google_drive_pkg.__name__,
|
||||
prompt_stem="domain_prompt",
|
||||
)
|
||||
|
|
@ -1,54 +0,0 @@
|
|||
You are the Google Drive operations sub-agent.
|
||||
You receive delegated instructions from a supervisor agent and return structured results for supervisor synthesis.
|
||||
|
||||
<goal>
|
||||
Execute Google Drive file operations accurately in the connected account.
|
||||
</goal>
|
||||
|
||||
<available_tools>
|
||||
- `create_google_drive_file`
|
||||
- `delete_google_drive_file`
|
||||
</available_tools>
|
||||
|
||||
<tool_policy>
|
||||
- Use only tools in `<available_tools>`.
|
||||
- Ensure target file identity/path is explicit before mutate actions.
|
||||
- If target is ambiguous, return `status=blocked` with candidate files.
|
||||
- Never invent file IDs/names or mutation outcomes.
|
||||
</tool_policy>
|
||||
|
||||
<out_of_scope>
|
||||
- Do not perform non-Google-Drive tasks.
|
||||
</out_of_scope>
|
||||
|
||||
<safety>
|
||||
- Never claim file mutation success without tool confirmation.
|
||||
</safety>
|
||||
|
||||
<failure_policy>
|
||||
- On tool failure, return `status=error` with concise recovery `next_step`.
|
||||
- On target ambiguity, return `status=blocked` with candidate files.
|
||||
</failure_policy>
|
||||
|
||||
<output_contract>
|
||||
Return **only** one JSON object (no markdown/prose):
|
||||
{
|
||||
"status": "success" | "partial" | "blocked" | "error",
|
||||
"action_summary": string,
|
||||
"evidence": {
|
||||
"file_id": string | null,
|
||||
"file_name": string | null,
|
||||
"operation": "create" | "delete" | null,
|
||||
"matched_candidates": [
|
||||
{ "file_id": string, "file_name": string | null }
|
||||
] | null
|
||||
},
|
||||
"next_step": string | null,
|
||||
"missing_fields": string[] | null,
|
||||
"assumptions": string[] | null
|
||||
}
|
||||
Rules:
|
||||
- `status=success` -> `next_step=null`, `missing_fields=null`.
|
||||
- `status=partial|blocked|error` -> `next_step` must be non-null.
|
||||
- `status=blocked` due to missing required inputs -> `missing_fields` must be non-null.
|
||||
</output_contract>
|
||||
|
|
@ -1,14 +0,0 @@
|
|||
"""Registry-backed Google Drive tools."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any
|
||||
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.registry import build_registry_tools_for_category
|
||||
|
||||
|
||||
def build_google_drive_tools(dependencies: dict[str, Any]) -> list[BaseTool]:
|
||||
"""Registry-backed tools for the ``google_drive`` category."""
|
||||
return build_registry_tools_for_category(dependencies, "google_drive")
|
||||
|
|
@ -1,9 +0,0 @@
|
|||
"""Luma vertical slice: registry tools, domain agent, ``domain_prompt.md``."""
|
||||
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.luma.agent import build_luma_domain_agent
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.luma.slice_tools import build_luma_tools
|
||||
|
||||
__all__ = [
|
||||
"build_luma_tools",
|
||||
"build_luma_domain_agent",
|
||||
]
|
||||
|
|
@ -1,21 +0,0 @@
|
|||
"""Luma domain agent graph."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from collections.abc import Sequence
|
||||
|
||||
import app.agents.multi_agent_chat.expert_agent.connectors.luma as luma_pkg
|
||||
from langchain_core.language_models import BaseChatModel
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.agents import build_domain_agent
|
||||
|
||||
|
||||
def build_luma_domain_agent(llm: BaseChatModel, tools: Sequence[BaseTool]):
|
||||
"""Compiled Luma domain-agent graph."""
|
||||
return build_domain_agent(
|
||||
llm,
|
||||
tools,
|
||||
prompt_package=luma_pkg.__name__,
|
||||
prompt_stem="domain_prompt",
|
||||
)
|
||||
|
|
@ -1,55 +0,0 @@
|
|||
You are the Luma operations sub-agent.
|
||||
You receive delegated instructions from a supervisor agent and return structured results for supervisor synthesis.
|
||||
|
||||
<goal>
|
||||
Execute Luma event listing, reads, and creation accurately.
|
||||
</goal>
|
||||
|
||||
<available_tools>
|
||||
- `list_luma_events`
|
||||
- `read_luma_event`
|
||||
- `create_luma_event`
|
||||
</available_tools>
|
||||
|
||||
<tool_policy>
|
||||
- Use only tools in `<available_tools>`.
|
||||
- Resolve relative dates against runtime timestamp.
|
||||
- If required event fields are missing, return `status=blocked` with `missing_fields`.
|
||||
- Never invent event IDs/times or creation outcomes.
|
||||
</tool_policy>
|
||||
|
||||
<out_of_scope>
|
||||
- Do not perform non-Luma tasks.
|
||||
</out_of_scope>
|
||||
|
||||
<safety>
|
||||
- Never claim event creation success without tool confirmation.
|
||||
</safety>
|
||||
|
||||
<failure_policy>
|
||||
- On tool failure, return `status=error` with concise recovery `next_step`.
|
||||
- On missing required fields, return `status=blocked` with `missing_fields`.
|
||||
</failure_policy>
|
||||
|
||||
<output_contract>
|
||||
Return **only** one JSON object (no markdown/prose):
|
||||
{
|
||||
"status": "success" | "partial" | "blocked" | "error",
|
||||
"action_summary": string,
|
||||
"evidence": {
|
||||
"event_id": string | null,
|
||||
"title": string | null,
|
||||
"start_at": string (ISO 8601 with timezone) | null,
|
||||
"matched_candidates": [
|
||||
{ "event_id": string, "title": string | null, "start_at": string | null }
|
||||
] | null
|
||||
},
|
||||
"next_step": string | null,
|
||||
"missing_fields": string[] | null,
|
||||
"assumptions": string[] | null
|
||||
}
|
||||
Rules:
|
||||
- `status=success` -> `next_step=null`, `missing_fields=null`.
|
||||
- `status=partial|blocked|error` -> `next_step` must be non-null.
|
||||
- `status=blocked` due to missing required inputs -> `missing_fields` must be non-null.
|
||||
</output_contract>
|
||||
|
|
@ -1,14 +0,0 @@
|
|||
"""Registry-backed Luma tools."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any
|
||||
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.registry import build_registry_tools_for_category
|
||||
|
||||
|
||||
def build_luma_tools(dependencies: dict[str, Any]) -> list[BaseTool]:
|
||||
"""Registry-backed tools for the ``luma`` category."""
|
||||
return build_registry_tools_for_category(dependencies, "luma")
|
||||
|
|
@ -1,11 +0,0 @@
|
|||
"""Notion connector slice."""
|
||||
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.notion.agent import build_notion_domain_agent
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.notion.slice_tools import (
|
||||
build_notion_tools,
|
||||
)
|
||||
|
||||
__all__ = [
|
||||
"build_notion_tools",
|
||||
"build_notion_domain_agent",
|
||||
]
|
||||
|
|
@ -1,21 +0,0 @@
|
|||
"""Notion domain agent graph."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from collections.abc import Sequence
|
||||
|
||||
import app.agents.multi_agent_chat.expert_agent.connectors.notion as notion_pkg
|
||||
from langchain_core.language_models import BaseChatModel
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.agents import build_domain_agent
|
||||
|
||||
|
||||
def build_notion_domain_agent(llm: BaseChatModel, tools: Sequence[BaseTool]):
|
||||
"""Compiled Notion domain-agent graph."""
|
||||
return build_domain_agent(
|
||||
llm,
|
||||
tools,
|
||||
prompt_package=notion_pkg.__name__,
|
||||
prompt_stem="domain_prompt",
|
||||
)
|
||||
|
|
@ -1,56 +0,0 @@
|
|||
You are the Notion operations sub-agent.
|
||||
You receive delegated instructions from a supervisor agent and return structured results for supervisor synthesis.
|
||||
|
||||
<goal>
|
||||
Execute Notion page operations accurately in the connected workspace.
|
||||
</goal>
|
||||
|
||||
<available_tools>
|
||||
- `create_notion_page`
|
||||
- `update_notion_page`
|
||||
- `delete_notion_page`
|
||||
</available_tools>
|
||||
|
||||
<tool_policy>
|
||||
- Use only tools in `<available_tools>`.
|
||||
- If target page context is unclear, do not ask the user directly; return `status=blocked` with candidate options and supervisor `next_step`.
|
||||
- Never invent page IDs, titles, or mutation outcomes.
|
||||
</tool_policy>
|
||||
|
||||
<out_of_scope>
|
||||
- Do not perform non-Notion tasks.
|
||||
</out_of_scope>
|
||||
|
||||
<safety>
|
||||
- Before update/delete, ensure the target page match is explicit.
|
||||
- Never claim mutation success without tool confirmation.
|
||||
</safety>
|
||||
|
||||
<failure_policy>
|
||||
- On tool failure, return `status=error` with concise retry/recovery `next_step`.
|
||||
- On ambiguous target, return `status=blocked` with candidate options.
|
||||
</failure_policy>
|
||||
|
||||
<output_contract>
|
||||
Return **only** one JSON object (no markdown/prose):
|
||||
{
|
||||
"status": "success" | "partial" | "blocked" | "error",
|
||||
"action_summary": string,
|
||||
"evidence": {
|
||||
"page_id": string | null,
|
||||
"page_title": string | null,
|
||||
"matched_candidates": [
|
||||
{ "page_id": string, "page_title": string | null }
|
||||
] | null
|
||||
},
|
||||
"next_step": string | null,
|
||||
"missing_fields": string[] | null,
|
||||
"assumptions": string[] | null
|
||||
}
|
||||
|
||||
Rules:
|
||||
- `status=success` -> `next_step=null`, `missing_fields=null`.
|
||||
- `status=partial|blocked|error` -> `next_step` must be non-null.
|
||||
- `status=blocked` due to missing required inputs -> `missing_fields` must be non-null.
|
||||
- On ambiguity, include candidate options in `evidence.matched_candidates`.
|
||||
</output_contract>
|
||||
|
|
@ -1,14 +0,0 @@
|
|||
"""Registry-backed Notion tools."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any
|
||||
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.registry import build_registry_tools_for_category
|
||||
|
||||
|
||||
def build_notion_tools(dependencies: dict[str, Any]) -> list[BaseTool]:
|
||||
"""Registry-backed tools for the ``notion`` category."""
|
||||
return build_registry_tools_for_category(dependencies, "notion")
|
||||
|
|
@ -1,11 +0,0 @@
|
|||
"""Microsoft OneDrive connector slice."""
|
||||
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.onedrive.agent import build_onedrive_domain_agent
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.onedrive.slice_tools import (
|
||||
build_onedrive_tools,
|
||||
)
|
||||
|
||||
__all__ = [
|
||||
"build_onedrive_tools",
|
||||
"build_onedrive_domain_agent",
|
||||
]
|
||||
|
|
@ -1,21 +0,0 @@
|
|||
"""Microsoft OneDrive domain agent graph."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from collections.abc import Sequence
|
||||
|
||||
import app.agents.multi_agent_chat.expert_agent.connectors.onedrive as onedrive_pkg
|
||||
from langchain_core.language_models import BaseChatModel
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.agents import build_domain_agent
|
||||
|
||||
|
||||
def build_onedrive_domain_agent(llm: BaseChatModel, tools: Sequence[BaseTool]):
|
||||
"""Compiled OneDrive domain-agent graph."""
|
||||
return build_domain_agent(
|
||||
llm,
|
||||
tools,
|
||||
prompt_package=onedrive_pkg.__name__,
|
||||
prompt_stem="domain_prompt",
|
||||
)
|
||||
|
|
@ -1,52 +0,0 @@
|
|||
You are the Microsoft OneDrive operations sub-agent.
|
||||
You receive delegated instructions from a supervisor agent and return structured results for supervisor synthesis.
|
||||
|
||||
<goal>
|
||||
Execute OneDrive file create/delete actions accurately in the connected account.
|
||||
</goal>
|
||||
|
||||
<available_tools>
|
||||
- `create_onedrive_file`
|
||||
- `delete_onedrive_file`
|
||||
</available_tools>
|
||||
|
||||
<tool_policy>
|
||||
- Use only tools in `<available_tools>`.
|
||||
- Ensure file identity/path is explicit before mutate actions.
|
||||
- If ambiguous, return `status=blocked` with candidate paths and supervisor next step.
|
||||
- Never invent IDs/paths or mutation results.
|
||||
</tool_policy>
|
||||
|
||||
<out_of_scope>
|
||||
- Do not perform non-OneDrive tasks.
|
||||
</out_of_scope>
|
||||
|
||||
<safety>
|
||||
- Never claim file mutation success without tool confirmation.
|
||||
</safety>
|
||||
|
||||
<failure_policy>
|
||||
- On tool failure, return `status=error` with concise recovery `next_step`.
|
||||
- On ambiguous targets, return `status=blocked` with candidate paths.
|
||||
</failure_policy>
|
||||
|
||||
<output_contract>
|
||||
Return **only** one JSON object (no markdown/prose):
|
||||
{
|
||||
"status": "success" | "partial" | "blocked" | "error",
|
||||
"action_summary": string,
|
||||
"evidence": {
|
||||
"file_id": string | null,
|
||||
"file_path": string | null,
|
||||
"operation": "create" | "delete" | null,
|
||||
"matched_candidates": string[] | null
|
||||
},
|
||||
"next_step": string | null,
|
||||
"missing_fields": string[] | null,
|
||||
"assumptions": string[] | null
|
||||
}
|
||||
Rules:
|
||||
- `status=success` -> `next_step=null`, `missing_fields=null`.
|
||||
- `status=partial|blocked|error` -> `next_step` must be non-null.
|
||||
- `status=blocked` due to missing required inputs -> `missing_fields` must be non-null.
|
||||
</output_contract>
|
||||
|
|
@ -1,14 +0,0 @@
|
|||
"""Registry-backed Microsoft OneDrive tools."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any
|
||||
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.registry import build_registry_tools_for_category
|
||||
|
||||
|
||||
def build_onedrive_tools(dependencies: dict[str, Any]) -> list[BaseTool]:
|
||||
"""Registry-backed tools for the ``onedrive`` category."""
|
||||
return build_registry_tools_for_category(dependencies, "onedrive")
|
||||
|
|
@ -1,9 +0,0 @@
|
|||
"""Microsoft Teams vertical slice: registry tools, domain agent, ``domain_prompt.md``."""
|
||||
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.teams.agent import build_teams_domain_agent
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.teams.slice_tools import build_teams_tools
|
||||
|
||||
__all__ = [
|
||||
"build_teams_tools",
|
||||
"build_teams_domain_agent",
|
||||
]
|
||||
|
|
@ -1,21 +0,0 @@
|
|||
"""Microsoft Teams domain agent graph."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from collections.abc import Sequence
|
||||
|
||||
import app.agents.multi_agent_chat.expert_agent.connectors.teams as teams_pkg
|
||||
from langchain_core.language_models import BaseChatModel
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.agents import build_domain_agent
|
||||
|
||||
|
||||
def build_teams_domain_agent(llm: BaseChatModel, tools: Sequence[BaseTool]):
|
||||
"""Compiled Microsoft Teams domain-agent graph."""
|
||||
return build_domain_agent(
|
||||
llm,
|
||||
tools,
|
||||
prompt_package=teams_pkg.__name__,
|
||||
prompt_stem="domain_prompt",
|
||||
)
|
||||
|
|
@ -1,55 +0,0 @@
|
|||
You are the Microsoft Teams operations sub-agent.
|
||||
You receive delegated instructions from a supervisor agent and return structured results for supervisor synthesis.
|
||||
|
||||
<goal>
|
||||
Execute Teams channel discovery, message reads, and sends accurately.
|
||||
</goal>
|
||||
|
||||
<available_tools>
|
||||
- `list_teams_channels`
|
||||
- `read_teams_messages`
|
||||
- `send_teams_message`
|
||||
</available_tools>
|
||||
|
||||
<tool_policy>
|
||||
- Use only tools in `<available_tools>`.
|
||||
- Resolve team/channel targets before read/send operations.
|
||||
- If ambiguous, return `status=blocked` with candidate channels and `next_step`.
|
||||
- Never invent message content, sender identity, timestamps, or delivery outcomes.
|
||||
</tool_policy>
|
||||
|
||||
<out_of_scope>
|
||||
- Do not perform non-Teams tasks.
|
||||
</out_of_scope>
|
||||
|
||||
<safety>
|
||||
- Never claim send success without tool confirmation.
|
||||
</safety>
|
||||
|
||||
<failure_policy>
|
||||
- On tool failure, return `status=error` with concise recovery `next_step`.
|
||||
- On unresolved destination ambiguity, return `status=blocked` with candidates.
|
||||
</failure_policy>
|
||||
|
||||
<output_contract>
|
||||
Return **only** one JSON object (no markdown/prose):
|
||||
{
|
||||
"status": "success" | "partial" | "blocked" | "error",
|
||||
"action_summary": string,
|
||||
"evidence": {
|
||||
"team_id": string | null,
|
||||
"channel_id": string | null,
|
||||
"message_id": string | null,
|
||||
"matched_candidates": [
|
||||
{ "team_id": string | null, "channel_id": string, "label": string | null }
|
||||
] | null
|
||||
},
|
||||
"next_step": string | null,
|
||||
"missing_fields": string[] | null,
|
||||
"assumptions": string[] | null
|
||||
}
|
||||
Rules:
|
||||
- `status=success` -> `next_step=null`, `missing_fields=null`.
|
||||
- `status=partial|blocked|error` -> `next_step` must be non-null.
|
||||
- `status=blocked` due to missing required inputs -> `missing_fields` must be non-null.
|
||||
</output_contract>
|
||||
|
|
@ -1,14 +0,0 @@
|
|||
"""Registry-backed Microsoft Teams tools."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any
|
||||
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.registry import build_registry_tools_for_category
|
||||
|
||||
|
||||
def build_teams_tools(dependencies: dict[str, Any]) -> list[BaseTool]:
|
||||
"""Registry-backed tools for the ``teams`` category."""
|
||||
return build_registry_tools_for_category(dependencies, "teams")
|
||||
|
|
@ -1,5 +0,0 @@
|
|||
"""Prompt-backed subgraphs for MCP OAuth integrations without a native tool registry slice."""
|
||||
|
||||
from app.agents.multi_agent_chat.expert_agent.mcp_bridge.agent import build_mcp_route_domain_agent
|
||||
|
||||
__all__ = ["build_mcp_route_domain_agent"]
|
||||
|
|
@ -1,25 +0,0 @@
|
|||
"""Domain agents for MCP-only OAuth integrations (no native registry slice)."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from collections.abc import Sequence
|
||||
|
||||
import app.agents.multi_agent_chat.expert_agent.mcp_bridge as mcp_bridge_pkg
|
||||
from langchain_core.language_models import BaseChatModel
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.agents import build_domain_agent
|
||||
|
||||
|
||||
def build_mcp_route_domain_agent(
|
||||
llm: BaseChatModel,
|
||||
route_key: str,
|
||||
tools: Sequence[BaseTool],
|
||||
):
|
||||
"""One subgraph per MCP-only route (``linear``, ``slack``, …); prompt stem ``{route_key}_domain``."""
|
||||
return build_domain_agent(
|
||||
llm,
|
||||
tools,
|
||||
prompt_package=mcp_bridge_pkg.__name__,
|
||||
prompt_stem=f"{route_key}_domain",
|
||||
)
|
||||
|
|
@ -1,46 +0,0 @@
|
|||
You are the Airtable MCP operations sub-agent.
|
||||
You receive delegated instructions from a supervisor agent and return structured results for supervisor synthesis.
|
||||
|
||||
<goal>
|
||||
Execute Airtable MCP base/table/record operations accurately.
|
||||
</goal>
|
||||
|
||||
<available_tools>
|
||||
- Runtime-provided Airtable MCP tools for bases, tables, and records.
|
||||
</available_tools>
|
||||
|
||||
<tool_policy>
|
||||
- Resolve base and table targets before record-level actions.
|
||||
- Do not guess IDs or schema fields.
|
||||
- If targets are ambiguous, return `status=blocked` with candidate options.
|
||||
- Never claim mutation success without tool confirmation.
|
||||
</tool_policy>
|
||||
|
||||
<out_of_scope>
|
||||
- Do not execute non-Airtable tasks.
|
||||
</out_of_scope>
|
||||
|
||||
<safety>
|
||||
- Never claim record mutations succeeded without tool confirmation.
|
||||
</safety>
|
||||
|
||||
<failure_policy>
|
||||
- On tool failure, return `status=error` with concise recovery `next_step`.
|
||||
- On unresolved target/schema ambiguity, return `status=blocked` with required options.
|
||||
</failure_policy>
|
||||
|
||||
<output_contract>
|
||||
Return **only** one JSON object (no markdown/prose):
|
||||
{
|
||||
"status": "success" | "partial" | "blocked" | "error",
|
||||
"action_summary": string,
|
||||
"evidence": { "items": object | null },
|
||||
"next_step": string | null,
|
||||
"missing_fields": string[] | null,
|
||||
"assumptions": string[] | null
|
||||
}
|
||||
Rules:
|
||||
- `status=success` -> `next_step=null`, `missing_fields=null`.
|
||||
- `status=partial|blocked|error` -> `next_step` must be non-null.
|
||||
- `status=blocked` due to missing required inputs -> `missing_fields` must be non-null.
|
||||
</output_contract>
|
||||
|
|
@ -1,45 +0,0 @@
|
|||
You are the ClickUp MCP operations sub-agent.
|
||||
You receive delegated instructions from a supervisor agent and return structured results for supervisor synthesis.
|
||||
|
||||
<goal>
|
||||
Execute ClickUp MCP operations accurately using only runtime-provided tools.
|
||||
</goal>
|
||||
|
||||
<available_tools>
|
||||
- Runtime-provided ClickUp MCP tools for task/workspace search and mutation.
|
||||
</available_tools>
|
||||
|
||||
<tool_policy>
|
||||
- Follow tool descriptions exactly.
|
||||
- If task/workspace target is ambiguous or missing, return `status=blocked` with required disambiguation fields.
|
||||
- Never claim mutation success without tool confirmation.
|
||||
</tool_policy>
|
||||
|
||||
<out_of_scope>
|
||||
- Do not execute non-ClickUp tasks.
|
||||
</out_of_scope>
|
||||
|
||||
<safety>
|
||||
- Never claim update/create success without tool confirmation.
|
||||
</safety>
|
||||
|
||||
<failure_policy>
|
||||
- On tool failure, return `status=error` with concise recovery `next_step`.
|
||||
- On unresolved ambiguity, return `status=blocked` with candidate options.
|
||||
</failure_policy>
|
||||
|
||||
<output_contract>
|
||||
Return **only** one JSON object (no markdown/prose):
|
||||
{
|
||||
"status": "success" | "partial" | "blocked" | "error",
|
||||
"action_summary": string,
|
||||
"evidence": { "items": object | null },
|
||||
"next_step": string | null,
|
||||
"missing_fields": string[] | null,
|
||||
"assumptions": string[] | null
|
||||
}
|
||||
Rules:
|
||||
- `status=success` -> `next_step=null`, `missing_fields=null`.
|
||||
- `status=partial|blocked|error` -> `next_step` must be non-null.
|
||||
- `status=blocked` due to missing required inputs -> `missing_fields` must be non-null.
|
||||
</output_contract>
|
||||
|
|
@ -1,46 +0,0 @@
|
|||
You are the generic MCP operations sub-agent for user-defined servers.
|
||||
You receive delegated instructions from a supervisor agent and return structured results for supervisor synthesis.
|
||||
|
||||
<goal>
|
||||
Execute tasks strictly through runtime-exposed MCP tools while respecting tool contracts.
|
||||
</goal>
|
||||
|
||||
<available_tools>
|
||||
- Runtime-provided MCP tools exposed by the connected custom server.
|
||||
</available_tools>
|
||||
|
||||
<tool_policy>
|
||||
- Follow each tool description and argument contract exactly.
|
||||
- Never assume a capability exists unless a tool explicitly provides it.
|
||||
- If required inputs are missing, return `status=blocked` with `missing_fields`.
|
||||
- Never claim success without tool output confirmation.
|
||||
</tool_policy>
|
||||
|
||||
<out_of_scope>
|
||||
- Do not claim capabilities that are not present in runtime-exposed tools.
|
||||
</out_of_scope>
|
||||
|
||||
<safety>
|
||||
- Never perform destructive operations without explicit delegated instruction and successful tool confirmation.
|
||||
</safety>
|
||||
|
||||
<failure_policy>
|
||||
- On tool failure, return `status=error` with concise recovery `next_step`.
|
||||
- On missing required inputs, return `status=blocked` with `missing_fields`.
|
||||
</failure_policy>
|
||||
|
||||
<output_contract>
|
||||
Return **only** one JSON object (no markdown/prose):
|
||||
{
|
||||
"status": "success" | "partial" | "blocked" | "error",
|
||||
"action_summary": string,
|
||||
"evidence": { "items": object | null },
|
||||
"next_step": string | null,
|
||||
"missing_fields": string[] | null,
|
||||
"assumptions": string[] | null
|
||||
}
|
||||
Rules:
|
||||
- `status=success` -> `next_step=null`, `missing_fields=null`.
|
||||
- `status=partial|blocked|error` -> `next_step` must be non-null.
|
||||
- `status=blocked` due to missing required inputs -> `missing_fields` must be non-null.
|
||||
</output_contract>
|
||||
|
|
@ -1,46 +0,0 @@
|
|||
You are the Jira MCP operations sub-agent.
|
||||
You receive delegated instructions from a supervisor agent and return structured results for supervisor synthesis.
|
||||
|
||||
<goal>
|
||||
Execute Jira MCP operations accurately, including discovery and issue mutation flows.
|
||||
</goal>
|
||||
|
||||
<available_tools>
|
||||
- Runtime-provided Jira MCP tools for site/project discovery, issue search, create, and update.
|
||||
</available_tools>
|
||||
|
||||
<tool_policy>
|
||||
- Respect discovery dependencies (site/project/issue-type) before mutate calls.
|
||||
- If required fields are missing or targets are ambiguous, return `status=blocked` with `missing_fields`.
|
||||
- Do not guess keys/IDs.
|
||||
- Never claim create/update success without tool confirmation.
|
||||
</tool_policy>
|
||||
|
||||
<out_of_scope>
|
||||
- Do not execute non-Jira tasks.
|
||||
</out_of_scope>
|
||||
|
||||
<safety>
|
||||
- Never perform destructive/mutating actions without explicit target resolution.
|
||||
</safety>
|
||||
|
||||
<failure_policy>
|
||||
- On tool failure, return `status=error` with concise recovery `next_step`.
|
||||
- On unresolved ambiguity, return `status=blocked` with candidates or missing fields.
|
||||
</failure_policy>
|
||||
|
||||
<output_contract>
|
||||
Return **only** one JSON object (no markdown/prose):
|
||||
{
|
||||
"status": "success" | "partial" | "blocked" | "error",
|
||||
"action_summary": string,
|
||||
"evidence": { "items": object | null },
|
||||
"next_step": string | null,
|
||||
"missing_fields": string[] | null,
|
||||
"assumptions": string[] | null
|
||||
}
|
||||
Rules:
|
||||
- `status=success` -> `next_step=null`, `missing_fields=null`.
|
||||
- `status=partial|blocked|error` -> `next_step` must be non-null.
|
||||
- `status=blocked` due to missing required inputs -> `missing_fields` must be non-null.
|
||||
</output_contract>
|
||||
|
|
@ -1,45 +0,0 @@
|
|||
You are the Linear MCP operations sub-agent.
|
||||
You receive delegated instructions from a supervisor agent and return structured results for supervisor synthesis.
|
||||
|
||||
<goal>
|
||||
Execute Linear MCP operations accurately using only available runtime tools.
|
||||
</goal>
|
||||
|
||||
<available_tools>
|
||||
- Runtime-provided Linear MCP tools for issues/projects/teams/workflows.
|
||||
</available_tools>
|
||||
|
||||
<tool_policy>
|
||||
- Follow tool descriptions exactly; do not assume unsupported endpoints.
|
||||
- If required identifiers or context are missing, return `status=blocked` with `missing_fields` and supervisor `next_step`.
|
||||
- Never invent IDs, statuses, or mutation outcomes.
|
||||
</tool_policy>
|
||||
|
||||
<out_of_scope>
|
||||
- Do not execute non-Linear tasks.
|
||||
</out_of_scope>
|
||||
|
||||
<safety>
|
||||
- Never claim mutation success without tool confirmation.
|
||||
</safety>
|
||||
|
||||
<failure_policy>
|
||||
- On tool failure, return `status=error` with concise recovery `next_step`.
|
||||
- On unresolved ambiguity, return `status=blocked` with candidates.
|
||||
</failure_policy>
|
||||
|
||||
<output_contract>
|
||||
Return **only** one JSON object (no markdown/prose):
|
||||
{
|
||||
"status": "success" | "partial" | "blocked" | "error",
|
||||
"action_summary": string,
|
||||
"evidence": { "items": object | null },
|
||||
"next_step": string | null,
|
||||
"missing_fields": string[] | null,
|
||||
"assumptions": string[] | null
|
||||
}
|
||||
Rules:
|
||||
- `status=success` -> `next_step=null`, `missing_fields=null`.
|
||||
- `status=partial|blocked|error` -> `next_step` must be non-null.
|
||||
- `status=blocked` due to missing required inputs -> `missing_fields` must be non-null.
|
||||
</output_contract>
|
||||
|
|
@ -1,45 +0,0 @@
|
|||
You are the Slack MCP operations sub-agent.
|
||||
You receive delegated instructions from a supervisor agent and return structured results for supervisor synthesis.
|
||||
|
||||
<goal>
|
||||
Execute Slack MCP reads/actions accurately in the connected workspace.
|
||||
</goal>
|
||||
|
||||
<available_tools>
|
||||
- Runtime-provided Slack MCP tools for search, channel/thread reads, and related actions.
|
||||
</available_tools>
|
||||
|
||||
<tool_policy>
|
||||
- Use only runtime-provided MCP tools and their documented arguments.
|
||||
- If channel/thread target is ambiguous, return `status=blocked` with candidate options.
|
||||
- Never invent message content, sender identity, timestamps, or delivery outcomes.
|
||||
</tool_policy>
|
||||
|
||||
<out_of_scope>
|
||||
- Do not execute non-Slack tasks.
|
||||
</out_of_scope>
|
||||
|
||||
<safety>
|
||||
- Never claim send/read success without tool evidence.
|
||||
</safety>
|
||||
|
||||
<failure_policy>
|
||||
- On tool failure, return `status=error` with concise recovery `next_step`.
|
||||
- On unresolved channel/thread ambiguity, return `status=blocked` with candidates.
|
||||
</failure_policy>
|
||||
|
||||
<output_contract>
|
||||
Return **only** one JSON object (no markdown/prose):
|
||||
{
|
||||
"status": "success" | "partial" | "blocked" | "error",
|
||||
"action_summary": string,
|
||||
"evidence": { "items": object | null },
|
||||
"next_step": string | null,
|
||||
"missing_fields": string[] | null,
|
||||
"assumptions": string[] | null
|
||||
}
|
||||
Rules:
|
||||
- `status=success` -> `next_step=null`, `missing_fields=null`.
|
||||
- `status=partial|blocked|error` -> `next_step` must be non-null.
|
||||
- `status=blocked` due to missing required inputs -> `missing_fields` must be non-null.
|
||||
</output_contract>
|
||||
|
|
@ -1,7 +0,0 @@
|
|||
"""Full-stack wiring (DB-scoped) on top of :mod:`routing` and :mod:`supervisor`."""
|
||||
|
||||
from app.agents.multi_agent_chat.integration.create_multi_agent_chat import (
|
||||
create_multi_agent_chat,
|
||||
)
|
||||
|
||||
__all__ = ["create_multi_agent_chat"]
|
||||
|
|
@ -1,256 +0,0 @@
|
|||
"""Build the multi-agent supervisor graph: MCP partition, registry, routing tools, optional SurfSense middleware."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
import asyncio
|
||||
import logging
|
||||
from typing import Any
|
||||
|
||||
from langchain_core.language_models import BaseChatModel
|
||||
from langchain_core.tools import BaseTool
|
||||
from langgraph.types import Checkpointer
|
||||
from sqlalchemy.ext.asyncio import AsyncSession
|
||||
|
||||
from app.agents.multi_agent_chat.core.mcp_partition import (
|
||||
fetch_mcp_connector_metadata_maps,
|
||||
partition_mcp_tools_by_expert_route,
|
||||
)
|
||||
from app.agents.multi_agent_chat.core.registry.dependencies import (
|
||||
build_registry_dependencies,
|
||||
coerce_thread_id_for_registry,
|
||||
)
|
||||
from app.agents.multi_agent_chat.middleware.supervisor_stack import (
|
||||
build_supervisor_middleware_stack,
|
||||
)
|
||||
from app.agents.multi_agent_chat.routing.supervisor_routing import (
|
||||
build_supervisor_routing_tools,
|
||||
)
|
||||
from app.agents.multi_agent_chat.supervisor import build_supervisor_agent
|
||||
from app.agents.new_chat.chat_deepagent import _map_connectors_to_searchable_types
|
||||
from app.agents.new_chat.context import SurfSenseContextSchema
|
||||
from app.agents.new_chat.feature_flags import get_flags
|
||||
from app.agents.new_chat.filesystem_backends import build_backend_resolver
|
||||
from app.agents.new_chat.filesystem_selection import FilesystemSelection
|
||||
from app.agents.new_chat.tools.mcp_tool import load_mcp_tools
|
||||
from app.db import ChatVisibility
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
|
||||
async def _discover_connectors_and_doc_types(
|
||||
*,
|
||||
connector_service: Any | None,
|
||||
search_space_id: int,
|
||||
available_connectors: list[str] | None,
|
||||
available_document_types: list[str] | None,
|
||||
) -> tuple[list[str] | None, list[str] | None]:
|
||||
"""Fill connector / document-type lists from ``connector_service`` when callers omit them."""
|
||||
connectors = available_connectors
|
||||
doc_types = available_document_types
|
||||
if connector_service is None:
|
||||
return connectors, doc_types
|
||||
try:
|
||||
if connectors is None:
|
||||
raw = await connector_service.get_available_connectors(search_space_id)
|
||||
if raw:
|
||||
connectors = _map_connectors_to_searchable_types(raw)
|
||||
if doc_types is None:
|
||||
doc_types = await connector_service.get_available_document_types(search_space_id)
|
||||
except Exception as exc:
|
||||
logger.warning("Failed to discover available connectors/document types: %s", exc)
|
||||
return connectors, doc_types
|
||||
|
||||
|
||||
async def _mcp_tools_by_expert_route(
|
||||
*,
|
||||
db_session: AsyncSession,
|
||||
search_space_id: int,
|
||||
) -> dict[str, list[BaseTool]] | None:
|
||||
mcp_flat = await load_mcp_tools(db_session, search_space_id)
|
||||
id_map, name_map = await fetch_mcp_connector_metadata_maps(db_session, search_space_id)
|
||||
return partition_mcp_tools_by_expert_route(mcp_flat, id_map, name_map)
|
||||
|
||||
|
||||
def _make_supervisor_routing_tools(
|
||||
llm: BaseChatModel,
|
||||
*,
|
||||
db_session: AsyncSession,
|
||||
search_space_id: int,
|
||||
user_id: str,
|
||||
thread_id: str | int | None,
|
||||
firecrawl_api_key: str | None,
|
||||
connector_service: Any | None,
|
||||
available_connectors: list[str] | None,
|
||||
available_document_types: list[str] | None,
|
||||
thread_visibility: ChatVisibility,
|
||||
mcp_tools_by_route: dict[str, list[BaseTool]] | None,
|
||||
) -> list[BaseTool]:
|
||||
registry_dependencies = build_registry_dependencies(
|
||||
db_session=db_session,
|
||||
search_space_id=search_space_id,
|
||||
user_id=user_id,
|
||||
thread_id=thread_id,
|
||||
llm=llm,
|
||||
firecrawl_api_key=firecrawl_api_key,
|
||||
connector_service=connector_service,
|
||||
available_connectors=available_connectors,
|
||||
available_document_types=available_document_types,
|
||||
thread_visibility=thread_visibility,
|
||||
)
|
||||
return build_supervisor_routing_tools(
|
||||
llm,
|
||||
registry_dependencies=registry_dependencies,
|
||||
include_deliverables=coerce_thread_id_for_registry(thread_id) is not None,
|
||||
mcp_tools_by_route=mcp_tools_by_route,
|
||||
available_connectors=available_connectors,
|
||||
thread_visibility=thread_visibility,
|
||||
)
|
||||
|
||||
|
||||
def _compile_supervisor_agent_sync(
|
||||
*,
|
||||
llm: BaseChatModel,
|
||||
routing_tools: list[BaseTool],
|
||||
checkpointer: Checkpointer | None,
|
||||
backend_resolver: Any,
|
||||
filesystem_mode: Any,
|
||||
search_space_id: int,
|
||||
user_id: str,
|
||||
thread_id: str | int | None,
|
||||
thread_visibility: ChatVisibility,
|
||||
anon_session_id: str | None,
|
||||
available_connectors: list[str] | None,
|
||||
available_document_types: list[str] | None,
|
||||
mentioned_document_ids: list[int] | None,
|
||||
max_input_tokens: int | None,
|
||||
citations_enabled: bool,
|
||||
) -> Any:
|
||||
"""CPU-heavy: middleware stack + ``create_agent`` (intended for ``asyncio.to_thread``)."""
|
||||
middleware = build_supervisor_middleware_stack(
|
||||
llm=llm,
|
||||
tools=routing_tools,
|
||||
backend_resolver=backend_resolver,
|
||||
filesystem_mode=filesystem_mode,
|
||||
search_space_id=search_space_id,
|
||||
user_id=user_id,
|
||||
thread_id=thread_id,
|
||||
visibility=thread_visibility,
|
||||
anon_session_id=anon_session_id,
|
||||
available_connectors=available_connectors,
|
||||
available_document_types=available_document_types,
|
||||
mentioned_document_ids=mentioned_document_ids,
|
||||
max_input_tokens=max_input_tokens,
|
||||
flags=get_flags(),
|
||||
)
|
||||
return build_supervisor_agent(
|
||||
llm,
|
||||
tools=routing_tools,
|
||||
checkpointer=checkpointer,
|
||||
thread_visibility=thread_visibility,
|
||||
middleware=middleware,
|
||||
context_schema=SurfSenseContextSchema,
|
||||
citations_enabled=citations_enabled,
|
||||
)
|
||||
|
||||
|
||||
async def create_multi_agent_chat(
|
||||
llm: BaseChatModel,
|
||||
*,
|
||||
db_session: AsyncSession,
|
||||
search_space_id: int,
|
||||
user_id: str,
|
||||
checkpointer: Checkpointer | None = None,
|
||||
thread_id: str | int | None = None,
|
||||
firecrawl_api_key: str | None = None,
|
||||
connector_service: Any | None = None,
|
||||
available_connectors: list[str] | None = None,
|
||||
available_document_types: list[str] | None = None,
|
||||
thread_visibility: ChatVisibility = ChatVisibility.PRIVATE,
|
||||
include_mcp_tools: bool = True,
|
||||
filesystem_selection: FilesystemSelection | None = None,
|
||||
anon_session_id: str | None = None,
|
||||
mentioned_document_ids: list[int] | None = None,
|
||||
max_input_tokens: int | None = None,
|
||||
surfsense_stack: bool = True,
|
||||
citations_enabled: bool | None = None,
|
||||
):
|
||||
"""Build the full multi-agent chat graph (supervisor + expert subgraphs via routing tools).
|
||||
|
||||
**Builtins** (:mod:`expert_agent.builtins`): registry-grouped **categories** (research, memory, deliverables).
|
||||
**Connectors** (:mod:`expert_agent.connectors`): **vendor integrations** — one subgraph per route in
|
||||
``TOOL_NAMES_BY_CATEGORY`` (e.g. calendar, confluence, discord, dropbox, gmail, google_drive, luma, notion, onedrive, teams).
|
||||
|
||||
MCP tools (via ``load_mcp_tools``) are partitioned inside this package and attached only
|
||||
to the matching expert subgraphs — not to the supervisor tool list as raw MCP calls. Inclusion matches
|
||||
``app.agents.new_chat.tools.registry.build_tools_async``: all tools returned by ``load_mcp_tools`` are merged
|
||||
after partitioning (no extra inventory filter on MCP). Connector routing uses ``available_connectors``:
|
||||
pass explicitly, or provide ``connector_service`` so lists are resolved like
|
||||
``create_surfsense_deep_agent`` (``get_available_connectors`` → searchable types).
|
||||
|
||||
Deliverables (thread-scoped reports, podcasts, etc.) are registered only when ``thread_id`` is set.
|
||||
|
||||
When ``surfsense_stack`` is true (default), the supervisor uses the same SurfSense middleware shell as
|
||||
the main single-agent chat (KB priority/tree, filesystem, compaction, permissions, etc.) except
|
||||
``SubAgentMiddleware`` / ``task``, since experts are separate graphs behind routing tools. Graph
|
||||
compilation runs in ``asyncio.to_thread`` so heavy CPU work does not block the event loop.
|
||||
|
||||
``citations_enabled``: when ``None``, defaults to ``True`` (same default as ``AgentConfig`` / main chat).
|
||||
"""
|
||||
citations = True if citations_enabled is None else citations_enabled
|
||||
connectors, doc_types = await _discover_connectors_and_doc_types(
|
||||
connector_service=connector_service,
|
||||
search_space_id=search_space_id,
|
||||
available_connectors=available_connectors,
|
||||
available_document_types=available_document_types,
|
||||
)
|
||||
|
||||
mcp_by_route: dict[str, list[BaseTool]] | None = None
|
||||
if include_mcp_tools:
|
||||
mcp_by_route = await _mcp_tools_by_expert_route(
|
||||
db_session=db_session, search_space_id=search_space_id
|
||||
)
|
||||
|
||||
routing_tools = _make_supervisor_routing_tools(
|
||||
llm,
|
||||
db_session=db_session,
|
||||
search_space_id=search_space_id,
|
||||
user_id=user_id,
|
||||
thread_id=thread_id,
|
||||
firecrawl_api_key=firecrawl_api_key,
|
||||
connector_service=connector_service,
|
||||
available_connectors=connectors,
|
||||
available_document_types=doc_types,
|
||||
thread_visibility=thread_visibility,
|
||||
mcp_tools_by_route=mcp_by_route,
|
||||
)
|
||||
|
||||
fs_sel = filesystem_selection or FilesystemSelection()
|
||||
backend_resolver = build_backend_resolver(fs_sel, search_space_id=search_space_id)
|
||||
|
||||
if not surfsense_stack:
|
||||
return build_supervisor_agent(
|
||||
llm,
|
||||
tools=routing_tools,
|
||||
checkpointer=checkpointer,
|
||||
thread_visibility=thread_visibility,
|
||||
citations_enabled=citations,
|
||||
)
|
||||
|
||||
return await asyncio.to_thread(
|
||||
_compile_supervisor_agent_sync,
|
||||
llm=llm,
|
||||
routing_tools=routing_tools,
|
||||
checkpointer=checkpointer,
|
||||
backend_resolver=backend_resolver,
|
||||
filesystem_mode=fs_sel.mode,
|
||||
search_space_id=search_space_id,
|
||||
user_id=user_id,
|
||||
thread_id=thread_id,
|
||||
thread_visibility=thread_visibility,
|
||||
anon_session_id=anon_session_id,
|
||||
available_connectors=connectors,
|
||||
available_document_types=doc_types,
|
||||
mentioned_document_ids=mentioned_document_ids,
|
||||
max_input_tokens=max_input_tokens,
|
||||
citations_enabled=citations,
|
||||
)
|
||||
|
|
@ -1,11 +0,0 @@
|
|||
"""SurfSense supervisor middleware (parity with the main single-agent chat, minus subagents)."""
|
||||
|
||||
from app.agents.multi_agent_chat.middleware.supervisor_stack import (
|
||||
build_supervisor_middleware_stack,
|
||||
parse_thread_id_for_action_log,
|
||||
)
|
||||
|
||||
__all__ = [
|
||||
"build_supervisor_middleware_stack",
|
||||
"parse_thread_id_for_action_log",
|
||||
]
|
||||
|
|
@ -1,363 +0,0 @@
|
|||
"""Supervisor middleware stack matching the main single-agent chat (no ``SubAgentMiddleware`` / ``task``)."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
import logging
|
||||
from collections.abc import Sequence
|
||||
from typing import Any
|
||||
|
||||
from deepagents.backends import StateBackend
|
||||
from deepagents.middleware.patch_tool_calls import PatchToolCallsMiddleware
|
||||
from deepagents.middleware.skills import SkillsMiddleware
|
||||
from langchain.agents.middleware import (
|
||||
LLMToolSelectorMiddleware,
|
||||
ModelCallLimitMiddleware,
|
||||
ModelFallbackMiddleware,
|
||||
TodoListMiddleware,
|
||||
ToolCallLimitMiddleware,
|
||||
)
|
||||
from langchain_anthropic.middleware import AnthropicPromptCachingMiddleware
|
||||
from langchain_core.language_models import BaseChatModel
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.new_chat.feature_flags import AgentFeatureFlags, get_flags
|
||||
from app.agents.new_chat.filesystem_selection import FilesystemMode
|
||||
from app.agents.new_chat.middleware import (
|
||||
ActionLogMiddleware,
|
||||
AnonymousDocumentMiddleware,
|
||||
BusyMutexMiddleware,
|
||||
ClearToolUsesEdit,
|
||||
DedupHITLToolCallsMiddleware,
|
||||
DoomLoopMiddleware,
|
||||
FileIntentMiddleware,
|
||||
KnowledgeBasePersistenceMiddleware,
|
||||
KnowledgePriorityMiddleware,
|
||||
KnowledgeTreeMiddleware,
|
||||
MemoryInjectionMiddleware,
|
||||
NoopInjectionMiddleware,
|
||||
OtelSpanMiddleware,
|
||||
RetryAfterMiddleware,
|
||||
SpillingContextEditingMiddleware,
|
||||
SpillToBackendEdit,
|
||||
SurfSenseFilesystemMiddleware,
|
||||
ToolCallNameRepairMiddleware,
|
||||
build_skills_backend_factory,
|
||||
create_surfsense_compaction_middleware,
|
||||
default_skills_sources,
|
||||
)
|
||||
from app.agents.new_chat.plugin_loader import (
|
||||
PluginContext,
|
||||
load_allowed_plugin_names_from_env,
|
||||
load_plugin_middlewares,
|
||||
)
|
||||
from app.agents.new_chat.tools.registry import BUILTIN_TOOLS
|
||||
from app.db import ChatVisibility
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
# Routing tools with heavy outputs — never prune via context editing when bound.
|
||||
_SUPERVISOR_PRUNE_PROTECTED: frozenset[str] = frozenset(
|
||||
{
|
||||
"deliverables",
|
||||
"invalid",
|
||||
# Align with single-agent surfacing of costly connector reads if names overlap later.
|
||||
"read_email",
|
||||
"search_emails",
|
||||
"generate_report",
|
||||
"generate_resume",
|
||||
"generate_podcast",
|
||||
"generate_video_presentation",
|
||||
"generate_image",
|
||||
}
|
||||
)
|
||||
|
||||
|
||||
def _safe_exclude_tools_supervisor(tools: Sequence[BaseTool]) -> tuple[str, ...]:
|
||||
enabled = {t.name for t in tools}
|
||||
return tuple(n for n in _SUPERVISOR_PRUNE_PROTECTED if n in enabled)
|
||||
|
||||
|
||||
def parse_thread_id_for_action_log(thread_id: int | str | None) -> int | None:
|
||||
"""Numeric DB thread ids only — UUID strings skip action logging (no FK row)."""
|
||||
if thread_id is None:
|
||||
return None
|
||||
if isinstance(thread_id, int):
|
||||
return thread_id
|
||||
s = str(thread_id).strip()
|
||||
if s.isdigit():
|
||||
return int(s)
|
||||
return None
|
||||
|
||||
|
||||
def build_supervisor_middleware_stack(
|
||||
*,
|
||||
llm: BaseChatModel,
|
||||
tools: Sequence[BaseTool],
|
||||
backend_resolver: Any,
|
||||
filesystem_mode: FilesystemMode,
|
||||
search_space_id: int,
|
||||
user_id: str | None,
|
||||
thread_id: int | str | None,
|
||||
visibility: ChatVisibility,
|
||||
anon_session_id: str | None,
|
||||
available_connectors: list[str] | None,
|
||||
available_document_types: list[str] | None,
|
||||
mentioned_document_ids: list[int] | None,
|
||||
max_input_tokens: int | None,
|
||||
flags: AgentFeatureFlags | None = None,
|
||||
) -> list[Any]:
|
||||
"""Build middleware list for the multi-agent supervisor (parity with ``_build_compiled_agent_blocking`` minus subagents)."""
|
||||
flags = flags or get_flags()
|
||||
|
||||
memory_middleware = MemoryInjectionMiddleware(
|
||||
user_id=user_id,
|
||||
search_space_id=search_space_id,
|
||||
thread_visibility=visibility,
|
||||
)
|
||||
|
||||
summarization_mw = create_surfsense_compaction_middleware(llm, StateBackend)
|
||||
_ = flags.enable_compaction_v2
|
||||
|
||||
context_edit_mw = None
|
||||
if (
|
||||
flags.enable_context_editing
|
||||
and not flags.disable_new_agent_stack
|
||||
and max_input_tokens
|
||||
):
|
||||
spill_edit = SpillToBackendEdit(
|
||||
trigger=int(max_input_tokens * 0.55),
|
||||
clear_at_least=int(max_input_tokens * 0.15),
|
||||
keep=5,
|
||||
exclude_tools=_safe_exclude_tools_supervisor(tools),
|
||||
clear_tool_inputs=True,
|
||||
)
|
||||
clear_edit = ClearToolUsesEdit(
|
||||
trigger=int(max_input_tokens * 0.55),
|
||||
clear_at_least=int(max_input_tokens * 0.15),
|
||||
keep=5,
|
||||
exclude_tools=_safe_exclude_tools_supervisor(tools),
|
||||
clear_tool_inputs=True,
|
||||
placeholder="[cleared - older tool output trimmed for context]",
|
||||
)
|
||||
context_edit_mw = SpillingContextEditingMiddleware(
|
||||
edits=[spill_edit, clear_edit],
|
||||
backend_resolver=backend_resolver,
|
||||
)
|
||||
|
||||
retry_mw = (
|
||||
RetryAfterMiddleware(max_retries=3)
|
||||
if flags.enable_retry_after and not flags.disable_new_agent_stack
|
||||
else None
|
||||
)
|
||||
fallback_mw: ModelFallbackMiddleware | None = None
|
||||
if flags.enable_model_fallback and not flags.disable_new_agent_stack:
|
||||
try:
|
||||
fallback_mw = ModelFallbackMiddleware(
|
||||
"openai:gpt-4o-mini",
|
||||
"anthropic:claude-3-5-haiku-20241022",
|
||||
)
|
||||
except Exception:
|
||||
logger.warning("ModelFallbackMiddleware init failed; skipping.")
|
||||
fallback_mw = None
|
||||
model_call_limit_mw = (
|
||||
ModelCallLimitMiddleware(
|
||||
thread_limit=120,
|
||||
run_limit=80,
|
||||
exit_behavior="end",
|
||||
)
|
||||
if flags.enable_model_call_limit and not flags.disable_new_agent_stack
|
||||
else None
|
||||
)
|
||||
tool_call_limit_mw = (
|
||||
ToolCallLimitMiddleware(
|
||||
thread_limit=300, run_limit=80, exit_behavior="continue"
|
||||
)
|
||||
if flags.enable_tool_call_limit and not flags.disable_new_agent_stack
|
||||
else None
|
||||
)
|
||||
|
||||
noop_mw = (
|
||||
NoopInjectionMiddleware()
|
||||
if flags.enable_compaction_v2 and not flags.disable_new_agent_stack
|
||||
else None
|
||||
)
|
||||
|
||||
repair_mw = None
|
||||
if flags.enable_tool_call_repair and not flags.disable_new_agent_stack:
|
||||
registered_names: set[str] = {t.name for t in tools}
|
||||
registered_names |= {
|
||||
"write_todos",
|
||||
"ls",
|
||||
"read_file",
|
||||
"write_file",
|
||||
"edit_file",
|
||||
"glob",
|
||||
"grep",
|
||||
"execute",
|
||||
# No ``task`` — multi-agent uses routing tools instead of SubAgentMiddleware.
|
||||
}
|
||||
repair_mw = ToolCallNameRepairMiddleware(
|
||||
registered_tool_names=registered_names,
|
||||
fuzzy_match_threshold=None,
|
||||
)
|
||||
|
||||
doom_loop_mw = (
|
||||
DoomLoopMiddleware(threshold=3)
|
||||
if flags.enable_doom_loop and not flags.disable_new_agent_stack
|
||||
else None
|
||||
)
|
||||
|
||||
thread_id_action_log = parse_thread_id_for_action_log(thread_id)
|
||||
action_log_mw: ActionLogMiddleware | None = None
|
||||
if (
|
||||
flags.enable_action_log
|
||||
and not flags.disable_new_agent_stack
|
||||
and thread_id_action_log is not None
|
||||
):
|
||||
try:
|
||||
tool_defs_by_name = {td.name: td for td in BUILTIN_TOOLS}
|
||||
action_log_mw = ActionLogMiddleware(
|
||||
thread_id=thread_id_action_log,
|
||||
search_space_id=search_space_id,
|
||||
user_id=user_id,
|
||||
tool_definitions=tool_defs_by_name,
|
||||
)
|
||||
except Exception: # pragma: no cover - defensive
|
||||
logger.warning(
|
||||
"ActionLogMiddleware init failed; running without it.",
|
||||
exc_info=True,
|
||||
)
|
||||
action_log_mw = None
|
||||
|
||||
busy_mutex_mw: BusyMutexMiddleware | None = (
|
||||
BusyMutexMiddleware()
|
||||
if flags.enable_busy_mutex and not flags.disable_new_agent_stack
|
||||
else None
|
||||
)
|
||||
|
||||
otel_mw: OtelSpanMiddleware | None = (
|
||||
OtelSpanMiddleware()
|
||||
if flags.enable_otel and not flags.disable_new_agent_stack
|
||||
else None
|
||||
)
|
||||
|
||||
plugin_middlewares: list[Any] = []
|
||||
if flags.enable_plugin_loader and not flags.disable_new_agent_stack:
|
||||
try:
|
||||
allowed_names = load_allowed_plugin_names_from_env()
|
||||
if allowed_names:
|
||||
plugin_middlewares = load_plugin_middlewares(
|
||||
PluginContext.build(
|
||||
search_space_id=search_space_id,
|
||||
user_id=user_id,
|
||||
thread_visibility=visibility,
|
||||
llm=llm,
|
||||
),
|
||||
allowed_plugin_names=allowed_names,
|
||||
)
|
||||
except Exception: # pragma: no cover - defensive
|
||||
logger.warning(
|
||||
"Plugin loader failed; continuing without plugins.",
|
||||
exc_info=True,
|
||||
)
|
||||
plugin_middlewares = []
|
||||
|
||||
skills_mw: SkillsMiddleware | None = None
|
||||
if flags.enable_skills and not flags.disable_new_agent_stack:
|
||||
try:
|
||||
skills_factory = build_skills_backend_factory(
|
||||
search_space_id=search_space_id
|
||||
if filesystem_mode == FilesystemMode.CLOUD
|
||||
else None,
|
||||
)
|
||||
skills_mw = SkillsMiddleware(
|
||||
backend=skills_factory,
|
||||
sources=default_skills_sources(),
|
||||
)
|
||||
except Exception as exc: # pragma: no cover - defensive
|
||||
logger.warning("SkillsMiddleware init failed; skipping: %s", exc)
|
||||
skills_mw = None
|
||||
|
||||
names = {t.name for t in tools}
|
||||
selector_mw: LLMToolSelectorMiddleware | None = None
|
||||
if (
|
||||
flags.enable_llm_tool_selector
|
||||
and not flags.disable_new_agent_stack
|
||||
and len(tools) > 30
|
||||
):
|
||||
try:
|
||||
selector_mw = LLMToolSelectorMiddleware(
|
||||
model="openai:gpt-4o-mini",
|
||||
max_tools=12,
|
||||
always_include=[
|
||||
n
|
||||
for n in (
|
||||
"research",
|
||||
"memory",
|
||||
"update_memory",
|
||||
"get_connected_accounts",
|
||||
"scrape_webpage",
|
||||
)
|
||||
if n in names
|
||||
],
|
||||
)
|
||||
except Exception:
|
||||
logger.warning("LLMToolSelectorMiddleware init failed; skipping.")
|
||||
selector_mw = None
|
||||
|
||||
deepagent_middleware = [
|
||||
busy_mutex_mw,
|
||||
otel_mw,
|
||||
TodoListMiddleware(),
|
||||
memory_middleware,
|
||||
AnonymousDocumentMiddleware(anon_session_id=anon_session_id)
|
||||
if filesystem_mode == FilesystemMode.CLOUD
|
||||
else None,
|
||||
KnowledgeTreeMiddleware(
|
||||
search_space_id=search_space_id,
|
||||
filesystem_mode=filesystem_mode,
|
||||
llm=llm,
|
||||
)
|
||||
if filesystem_mode == FilesystemMode.CLOUD
|
||||
else None,
|
||||
KnowledgePriorityMiddleware(
|
||||
llm=llm,
|
||||
search_space_id=search_space_id,
|
||||
filesystem_mode=filesystem_mode,
|
||||
available_connectors=available_connectors,
|
||||
available_document_types=available_document_types,
|
||||
mentioned_document_ids=mentioned_document_ids,
|
||||
),
|
||||
FileIntentMiddleware(llm=llm),
|
||||
SurfSenseFilesystemMiddleware(
|
||||
backend=backend_resolver,
|
||||
filesystem_mode=filesystem_mode,
|
||||
search_space_id=search_space_id,
|
||||
created_by_id=user_id,
|
||||
thread_id=thread_id,
|
||||
),
|
||||
KnowledgeBasePersistenceMiddleware(
|
||||
search_space_id=search_space_id,
|
||||
created_by_id=user_id,
|
||||
filesystem_mode=filesystem_mode,
|
||||
)
|
||||
if filesystem_mode == FilesystemMode.CLOUD
|
||||
else None,
|
||||
skills_mw,
|
||||
selector_mw,
|
||||
model_call_limit_mw,
|
||||
tool_call_limit_mw,
|
||||
context_edit_mw,
|
||||
summarization_mw,
|
||||
noop_mw,
|
||||
retry_mw,
|
||||
fallback_mw,
|
||||
repair_mw,
|
||||
doom_loop_mw,
|
||||
action_log_mw,
|
||||
PatchToolCallsMiddleware(),
|
||||
DedupHITLToolCallsMiddleware(agent_tools=list(tools)),
|
||||
*plugin_middlewares,
|
||||
AnthropicPromptCachingMiddleware(unsupported_model_behavior="ignore"),
|
||||
]
|
||||
return [m for m in deepagent_middleware if m is not None]
|
||||
|
|
@ -1,13 +0,0 @@
|
|||
"""Supervisor routing: domain-agent wrappers and composed routing tool lists."""
|
||||
|
||||
from app.agents.multi_agent_chat.routing.domain_routing_spec import DomainRoutingSpec
|
||||
from app.agents.multi_agent_chat.routing.from_domain_agents import routing_tools_from_specs
|
||||
from app.agents.multi_agent_chat.routing.supervisor_routing import build_supervisor_routing_tools
|
||||
from app.agents.multi_agent_chat.core.invocation import extract_last_assistant_text
|
||||
|
||||
__all__ = [
|
||||
"DomainRoutingSpec",
|
||||
"build_supervisor_routing_tools",
|
||||
"extract_last_assistant_text",
|
||||
"routing_tools_from_specs",
|
||||
]
|
||||
|
|
@ -1,22 +0,0 @@
|
|||
"""Declarative description of one supervisor routing tool → domain agent."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from collections.abc import Callable
|
||||
from dataclasses import dataclass
|
||||
from typing import Any
|
||||
|
||||
|
||||
@dataclass(frozen=True)
|
||||
class DomainRoutingSpec:
|
||||
"""One supervisor-facing routing ``@tool`` bound to a compiled domain graph.
|
||||
|
||||
``curated_context`` is optional for **any** route: when set, the routing tool prepends its return
|
||||
value into the child task via :func:`~app.agents.multi_agent_chat.core.delegation.compose_child_task`.
|
||||
:func:`build_supervisor_routing_tools` does not pass it (all routes treated the same); use when building specs manually.
|
||||
"""
|
||||
|
||||
tool_name: str
|
||||
description: str
|
||||
domain_agent: Any
|
||||
curated_context: Callable[[str], str | None] | None = None
|
||||
|
|
@ -1,123 +0,0 @@
|
|||
"""LangChain ``@tool`` wrappers that invoke compiled domain-agent graphs (supervisor-facing only)."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from collections.abc import Sequence
|
||||
import json
|
||||
from typing import Any
|
||||
|
||||
from langchain_core.tools import BaseTool, tool
|
||||
|
||||
from app.agents.multi_agent_chat.core.delegation import compose_child_task
|
||||
from app.agents.multi_agent_chat.core.invocation import extract_last_assistant_text
|
||||
from app.agents.multi_agent_chat.routing.domain_routing_spec import DomainRoutingSpec
|
||||
|
||||
_ALLOWED_STATUSES = {"success", "partial", "blocked", "error"}
|
||||
_REQUIRED_KEYS = {
|
||||
"status",
|
||||
"action_summary",
|
||||
"evidence",
|
||||
"next_step",
|
||||
"missing_fields",
|
||||
"assumptions",
|
||||
}
|
||||
|
||||
|
||||
def _fallback_payload(spec: DomainRoutingSpec, reason: str, raw_text: str) -> dict[str, Any]:
|
||||
preview = raw_text[:800]
|
||||
return {
|
||||
"status": "error",
|
||||
"action_summary": "Domain agent output failed JSON contract validation.",
|
||||
"evidence": {
|
||||
"route_tool": spec.tool_name,
|
||||
"validation_error": reason,
|
||||
"raw_output_preview": preview,
|
||||
},
|
||||
"next_step": (
|
||||
"Re-delegate with a strict reminder to return exactly one JSON object "
|
||||
"matching the output_contract."
|
||||
),
|
||||
"missing_fields": None,
|
||||
"assumptions": None,
|
||||
}
|
||||
|
||||
|
||||
def _validate_contract_payload(payload: dict[str, Any]) -> str | None:
|
||||
missing = sorted(_REQUIRED_KEYS - set(payload))
|
||||
if missing:
|
||||
return f"missing required keys: {', '.join(missing)}"
|
||||
|
||||
status = payload.get("status")
|
||||
if status not in _ALLOWED_STATUSES:
|
||||
return "invalid status value"
|
||||
|
||||
action_summary = payload.get("action_summary")
|
||||
if not isinstance(action_summary, str) or not action_summary.strip():
|
||||
return "action_summary must be a non-empty string"
|
||||
|
||||
evidence = payload.get("evidence")
|
||||
if not isinstance(evidence, dict):
|
||||
return "evidence must be an object"
|
||||
|
||||
next_step = payload.get("next_step")
|
||||
if status == "success":
|
||||
if next_step is not None:
|
||||
return "next_step must be null when status=success"
|
||||
if payload.get("missing_fields") is not None:
|
||||
return "missing_fields must be null when status=success"
|
||||
else:
|
||||
if not isinstance(next_step, str) or not next_step.strip():
|
||||
return "next_step must be a non-empty string for non-success statuses"
|
||||
|
||||
missing_fields = payload.get("missing_fields")
|
||||
if missing_fields is not None:
|
||||
if not isinstance(missing_fields, list) or any(
|
||||
not isinstance(item, str) or not item.strip() for item in missing_fields
|
||||
):
|
||||
return "missing_fields must be null or a list of non-empty strings"
|
||||
|
||||
assumptions = payload.get("assumptions")
|
||||
if assumptions is not None:
|
||||
if not isinstance(assumptions, list) or any(
|
||||
not isinstance(item, str) or not item.strip() for item in assumptions
|
||||
):
|
||||
return "assumptions must be null or a list of non-empty strings"
|
||||
|
||||
return None
|
||||
|
||||
|
||||
def _normalize_domain_output(spec: DomainRoutingSpec, raw_text: str) -> str:
|
||||
try:
|
||||
parsed = json.loads(raw_text)
|
||||
except json.JSONDecodeError as exc:
|
||||
fallback = _fallback_payload(spec, f"invalid JSON: {exc.msg}", raw_text)
|
||||
return json.dumps(fallback)
|
||||
|
||||
if not isinstance(parsed, dict):
|
||||
fallback = _fallback_payload(spec, "top-level JSON must be an object", raw_text)
|
||||
return json.dumps(fallback)
|
||||
|
||||
validation_error = _validate_contract_payload(parsed)
|
||||
if validation_error:
|
||||
fallback = _fallback_payload(spec, validation_error, raw_text)
|
||||
return json.dumps(fallback)
|
||||
|
||||
return json.dumps(parsed)
|
||||
|
||||
|
||||
def _routing_tool_for_spec(spec: DomainRoutingSpec) -> BaseTool:
|
||||
@tool(spec.tool_name, description=spec.description)
|
||||
async def _route(task: str) -> str:
|
||||
curated = spec.curated_context(task) if spec.curated_context else None
|
||||
content = compose_child_task(task, curated_context=curated)
|
||||
result = await spec.domain_agent.ainvoke(
|
||||
{"messages": [{"role": "user", "content": content}]},
|
||||
)
|
||||
return _normalize_domain_output(spec, extract_last_assistant_text(result))
|
||||
|
||||
return _route
|
||||
|
||||
|
||||
def routing_tools_from_specs(specs: Sequence[DomainRoutingSpec]) -> list[BaseTool]:
|
||||
"""Build one supervisor-facing routing ``@tool`` per :class:`DomainRoutingSpec`."""
|
||||
return [_routing_tool_for_spec(spec) for spec in specs]
|
||||
|
|
@ -1,57 +0,0 @@
|
|||
"""Gate supervisor routing tools by connected searchable connector types (aligned with main chat KB).
|
||||
|
||||
When ``available_connectors`` is ``None``, all routes are emitted (caller did not pass an inventory).
|
||||
|
||||
When provided, a connector route is emitted only if at least one required searchable type is present.
|
||||
MCP tools are filtered upstream in :func:`~app.agents.multi_agent_chat.core.mcp_partition.partition_mcp_tools_by_expert_route`
|
||||
so merges only include tools for connected accounts.
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
# Route tool_name → searchable connector / doc-type strings (same family as
|
||||
# ``chat_deepagent._CONNECTOR_TYPE_TO_SEARCHABLE`` values in ``available_connectors``).
|
||||
_ROUTE_REQUIRES_ANY: dict[str, frozenset[str]] = {
|
||||
"calendar": frozenset(
|
||||
{"GOOGLE_CALENDAR_CONNECTOR", "COMPOSIO_GOOGLE_CALENDAR_CONNECTOR"}
|
||||
),
|
||||
"confluence": frozenset({"CONFLUENCE_CONNECTOR"}),
|
||||
"discord": frozenset({"DISCORD_CONNECTOR"}),
|
||||
"dropbox": frozenset({"DROPBOX_FILE"}),
|
||||
"gmail": frozenset({"GOOGLE_GMAIL_CONNECTOR", "COMPOSIO_GMAIL_CONNECTOR"}),
|
||||
"google_drive": frozenset(
|
||||
{"GOOGLE_DRIVE_FILE", "COMPOSIO_GOOGLE_DRIVE_CONNECTOR"}
|
||||
),
|
||||
"luma": frozenset({"LUMA_CONNECTOR"}),
|
||||
"notion": frozenset({"NOTION_CONNECTOR"}),
|
||||
"onedrive": frozenset({"ONEDRIVE_FILE"}),
|
||||
"teams": frozenset({"TEAMS_CONNECTOR"}),
|
||||
# MCP-only supervisor routes (see ``core.mcp_partition.MCP_ONLY_ROUTE_KEYS_IN_ORDER``).
|
||||
"linear": frozenset({"LINEAR_CONNECTOR"}),
|
||||
"slack": frozenset({"SLACK_CONNECTOR"}),
|
||||
"jira": frozenset({"JIRA_CONNECTOR"}),
|
||||
"clickup": frozenset({"CLICKUP_CONNECTOR"}),
|
||||
"airtable": frozenset({"AIRTABLE_CONNECTOR"}),
|
||||
# generic_mcp route intentionally disabled for now.
|
||||
# "generic_mcp": frozenset({"MCP_CONNECTOR"}),
|
||||
}
|
||||
|
||||
|
||||
def include_connector_route(
|
||||
route_key: str,
|
||||
available_connectors: list[str] | None,
|
||||
) -> bool:
|
||||
"""Return whether to register this connector route on the supervisor.
|
||||
|
||||
If ``available_connectors`` is omitted, preserve legacy behaviour (emit the route).
|
||||
|
||||
Otherwise require at least one matching entry in ``available_connectors`` for connector-backed routes.
|
||||
Builtin routes (research, memory, …) have no entry in ``_ROUTE_REQUIRES_ANY`` and are always included.
|
||||
"""
|
||||
if available_connectors is None:
|
||||
return True
|
||||
required = _ROUTE_REQUIRES_ANY.get(route_key)
|
||||
if required is None:
|
||||
return True
|
||||
avail = set(available_connectors)
|
||||
return bool(required & avail)
|
||||
|
|
@ -1,327 +0,0 @@
|
|||
"""Compose domain agents + tool lists into supervisor routing tools (one ``@tool`` per category)."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any
|
||||
|
||||
from langchain_core.language_models import BaseChatModel
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
from app.agents.multi_agent_chat.core.mcp_partition import MCP_ONLY_ROUTE_KEYS_IN_ORDER
|
||||
from app.agents.multi_agent_chat.expert_agent.builtins.deliverables import (
|
||||
build_deliverables_domain_agent,
|
||||
build_deliverables_tools,
|
||||
)
|
||||
from app.agents.multi_agent_chat.expert_agent.builtins.memory import (
|
||||
build_memory_domain_agent,
|
||||
build_memory_tools,
|
||||
)
|
||||
from app.agents.multi_agent_chat.expert_agent.builtins.research import (
|
||||
build_research_domain_agent,
|
||||
build_research_tools,
|
||||
)
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.calendar import (
|
||||
build_calendar_domain_agent,
|
||||
build_calendar_tools,
|
||||
)
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.confluence import (
|
||||
build_confluence_domain_agent,
|
||||
build_confluence_tools,
|
||||
)
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.discord import (
|
||||
build_discord_domain_agent,
|
||||
build_discord_tools,
|
||||
)
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.dropbox import (
|
||||
build_dropbox_domain_agent,
|
||||
build_dropbox_tools,
|
||||
)
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.gmail import (
|
||||
build_gmail_domain_agent,
|
||||
build_gmail_tools,
|
||||
)
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.google_drive import (
|
||||
build_google_drive_domain_agent,
|
||||
build_google_drive_tools,
|
||||
)
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.luma import (
|
||||
build_luma_domain_agent,
|
||||
build_luma_tools,
|
||||
)
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.notion import (
|
||||
build_notion_domain_agent,
|
||||
build_notion_tools,
|
||||
)
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.onedrive import (
|
||||
build_onedrive_domain_agent,
|
||||
build_onedrive_tools,
|
||||
)
|
||||
from app.agents.multi_agent_chat.expert_agent.connectors.teams import (
|
||||
build_teams_domain_agent,
|
||||
build_teams_tools,
|
||||
)
|
||||
from app.agents.multi_agent_chat.expert_agent.mcp_bridge import (
|
||||
build_mcp_route_domain_agent,
|
||||
)
|
||||
from app.agents.multi_agent_chat.routing.domain_routing_spec import DomainRoutingSpec
|
||||
from app.agents.multi_agent_chat.routing.from_domain_agents import (
|
||||
routing_tools_from_specs,
|
||||
)
|
||||
from app.agents.multi_agent_chat.routing.route_connector_gate import (
|
||||
include_connector_route,
|
||||
)
|
||||
from app.db import ChatVisibility
|
||||
|
||||
_MCP_ONLY_ROUTE_DESCRIPTIONS: dict[str, str] = {
|
||||
"linear": (
|
||||
"Use for Linear issue/project work: find/create issues, update status/assignees, review project progress, and inspect cycles."
|
||||
),
|
||||
"slack": (
|
||||
"Use for Slack channel communication: read channel/thread history, summarize conversations, and post replies."
|
||||
),
|
||||
"jira": (
|
||||
"Use for Jira issue/project workflows: search issues, inspect fields, update tickets, and move work through workflow states."
|
||||
),
|
||||
"clickup": (
|
||||
"Use for ClickUp task management: find tasks/lists, update task fields, and track execution progress."
|
||||
),
|
||||
"airtable": (
|
||||
"Use for Airtable structured data operations: locate bases/tables and create/read/update records."
|
||||
),
|
||||
# generic_mcp intentionally disabled for now.
|
||||
# "generic_mcp": (
|
||||
# "Use as a fallback for custom connected app tasks not covered by a named specialist. "
|
||||
# "Do not use if another specialist clearly matches."
|
||||
# ),
|
||||
}
|
||||
|
||||
|
||||
def _memory_route_description(thread_visibility: ChatVisibility | None) -> str:
|
||||
if thread_visibility == ChatVisibility.SEARCH_SPACE:
|
||||
return "Use for storing durable team memory: shared team preferences, conventions, and long-lived team facts."
|
||||
return "Use for storing durable user memory: personal preferences, instructions, and long-lived user facts."
|
||||
|
||||
|
||||
def build_supervisor_routing_tools(
|
||||
llm: BaseChatModel,
|
||||
*,
|
||||
registry_dependencies: dict[str, Any] | None = None,
|
||||
include_deliverables: bool = True,
|
||||
mcp_tools_by_route: dict[str, list[BaseTool]] | None = None,
|
||||
available_connectors: list[str] | None = None,
|
||||
thread_visibility: ChatVisibility | None = None,
|
||||
) -> list[BaseTool]:
|
||||
"""Build supervisor routing tools: builtins first, then connector experts (same pattern for all).
|
||||
|
||||
Requires ``registry_dependencies`` to produce any routing tools; otherwise returns an empty list.
|
||||
|
||||
Pass ``registry_dependencies`` from
|
||||
:func:`app.agents.multi_agent_chat.core.registry.build_registry_dependencies`
|
||||
for builtins (**research**, **memory**, **deliverables** when ``include_deliverables``) and every
|
||||
registry-backed connector route.
|
||||
|
||||
``mcp_tools_by_route`` maps route keys to MCP tools merged into the matching expert subgraph.
|
||||
|
||||
When ``available_connectors`` is set (searchable connector strings, same shape as the main chat agent),
|
||||
a connector-backed route is registered only if its required searchable connector type is available.
|
||||
"""
|
||||
if registry_dependencies is None:
|
||||
return routing_tools_from_specs([])
|
||||
|
||||
mcp = mcp_tools_by_route or {}
|
||||
specs: list[DomainRoutingSpec] = []
|
||||
|
||||
research_tools = build_research_tools(registry_dependencies)
|
||||
research_agent = build_research_domain_agent(llm, research_tools)
|
||||
specs.append(
|
||||
DomainRoutingSpec(
|
||||
tool_name="research",
|
||||
description=(
|
||||
"Use for external research: find sources on the web, extract evidence, and answer documentation questions."
|
||||
),
|
||||
domain_agent=research_agent,
|
||||
),
|
||||
)
|
||||
|
||||
memory_tools = build_memory_tools(registry_dependencies)
|
||||
memory_agent = build_memory_domain_agent(
|
||||
llm,
|
||||
memory_tools,
|
||||
thread_visibility=thread_visibility,
|
||||
)
|
||||
specs.append(
|
||||
DomainRoutingSpec(
|
||||
tool_name="memory",
|
||||
description=_memory_route_description(thread_visibility),
|
||||
domain_agent=memory_agent,
|
||||
),
|
||||
)
|
||||
|
||||
if include_deliverables:
|
||||
deliverables_tools = build_deliverables_tools(registry_dependencies)
|
||||
deliverables_agent = build_deliverables_domain_agent(llm, deliverables_tools)
|
||||
specs.append(
|
||||
DomainRoutingSpec(
|
||||
tool_name="deliverables",
|
||||
description=(
|
||||
"Use for deliverables and shareable artifacts: generated reports, podcasts, "
|
||||
"video presentations, resumes, and images—not for routine lookups or single small edits elsewhere."
|
||||
),
|
||||
domain_agent=deliverables_agent,
|
||||
),
|
||||
)
|
||||
|
||||
# Connector experts (registry-backed + optional MCP merge): alphabetical by route key.
|
||||
if include_connector_route("calendar", available_connectors):
|
||||
calendar_agent = build_calendar_domain_agent(
|
||||
llm,
|
||||
build_calendar_tools(registry_dependencies) + mcp.get("calendar", []),
|
||||
)
|
||||
specs.append(
|
||||
DomainRoutingSpec(
|
||||
tool_name="calendar",
|
||||
description=(
|
||||
"Use for calendar planning and scheduling: check availability, read event details, create events, and update events."
|
||||
),
|
||||
domain_agent=calendar_agent,
|
||||
),
|
||||
)
|
||||
|
||||
if include_connector_route("confluence", available_connectors):
|
||||
confluence_tools = build_confluence_tools(registry_dependencies)
|
||||
confluence_agent = build_confluence_domain_agent(llm, confluence_tools)
|
||||
specs.append(
|
||||
DomainRoutingSpec(
|
||||
tool_name="confluence",
|
||||
description=(
|
||||
"Use for Confluence knowledge pages: search/read existing pages, create new pages, and update page content."
|
||||
),
|
||||
domain_agent=confluence_agent,
|
||||
),
|
||||
)
|
||||
|
||||
if include_connector_route("discord", available_connectors):
|
||||
discord_tools = build_discord_tools(registry_dependencies)
|
||||
discord_agent = build_discord_domain_agent(llm, discord_tools + mcp.get("discord", []))
|
||||
specs.append(
|
||||
DomainRoutingSpec(
|
||||
tool_name="discord",
|
||||
description=(
|
||||
"Use for Discord communication: read channel/thread messages, gather context, and send replies."
|
||||
),
|
||||
domain_agent=discord_agent,
|
||||
),
|
||||
)
|
||||
|
||||
if include_connector_route("dropbox", available_connectors):
|
||||
dropbox_tools = build_dropbox_tools(registry_dependencies)
|
||||
dropbox_agent = build_dropbox_domain_agent(llm, dropbox_tools)
|
||||
specs.append(
|
||||
DomainRoutingSpec(
|
||||
tool_name="dropbox",
|
||||
description=(
|
||||
"Use for Dropbox file storage tasks: browse folders, read files, and manage Dropbox file content."
|
||||
),
|
||||
domain_agent=dropbox_agent,
|
||||
),
|
||||
)
|
||||
|
||||
if include_connector_route("gmail", available_connectors):
|
||||
gmail_agent = build_gmail_domain_agent(
|
||||
llm,
|
||||
build_gmail_tools(registry_dependencies) + mcp.get("gmail", []),
|
||||
)
|
||||
specs.append(
|
||||
DomainRoutingSpec(
|
||||
tool_name="gmail",
|
||||
description=(
|
||||
"Use for Gmail inbox actions: search/read emails, draft or update replies, send messages, and trash emails."
|
||||
),
|
||||
domain_agent=gmail_agent,
|
||||
),
|
||||
)
|
||||
|
||||
if include_connector_route("google_drive", available_connectors):
|
||||
google_drive_tools = build_google_drive_tools(registry_dependencies)
|
||||
google_drive_agent = build_google_drive_domain_agent(llm, google_drive_tools)
|
||||
specs.append(
|
||||
DomainRoutingSpec(
|
||||
tool_name="google_drive",
|
||||
description=(
|
||||
"Use for Google Drive document/file tasks: locate files, inspect content, and manage Drive files or folders."
|
||||
),
|
||||
domain_agent=google_drive_agent,
|
||||
),
|
||||
)
|
||||
|
||||
if include_connector_route("luma", available_connectors):
|
||||
luma_tools = build_luma_tools(registry_dependencies)
|
||||
luma_agent = build_luma_domain_agent(llm, luma_tools + mcp.get("luma", []))
|
||||
specs.append(
|
||||
DomainRoutingSpec(
|
||||
tool_name="luma",
|
||||
description=(
|
||||
"Use for Luma event operations: list events, inspect event details, and create new events."
|
||||
),
|
||||
domain_agent=luma_agent,
|
||||
),
|
||||
)
|
||||
|
||||
if include_connector_route("notion", available_connectors):
|
||||
notion_tools = build_notion_tools(registry_dependencies)
|
||||
notion_agent = build_notion_domain_agent(llm, notion_tools)
|
||||
specs.append(
|
||||
DomainRoutingSpec(
|
||||
tool_name="notion",
|
||||
description=(
|
||||
"Use for Notion workspace pages: create pages, update page content, and delete pages."
|
||||
),
|
||||
domain_agent=notion_agent,
|
||||
),
|
||||
)
|
||||
|
||||
if include_connector_route("onedrive", available_connectors):
|
||||
onedrive_tools = build_onedrive_tools(registry_dependencies)
|
||||
onedrive_agent = build_onedrive_domain_agent(llm, onedrive_tools)
|
||||
specs.append(
|
||||
DomainRoutingSpec(
|
||||
tool_name="onedrive",
|
||||
description=(
|
||||
"Use for OneDrive file storage tasks: browse folders, read files, and manage OneDrive file content."
|
||||
),
|
||||
domain_agent=onedrive_agent,
|
||||
),
|
||||
)
|
||||
|
||||
if include_connector_route("teams", available_connectors):
|
||||
teams_tools = build_teams_tools(registry_dependencies)
|
||||
teams_agent = build_teams_domain_agent(llm, teams_tools + mcp.get("teams", []))
|
||||
specs.append(
|
||||
DomainRoutingSpec(
|
||||
tool_name="teams",
|
||||
description=(
|
||||
"Use for Microsoft Teams communication: read channel/thread messages, gather context, and post replies."
|
||||
),
|
||||
domain_agent=teams_agent,
|
||||
),
|
||||
)
|
||||
|
||||
for route_key in MCP_ONLY_ROUTE_KEYS_IN_ORDER:
|
||||
only_mcp = mcp.get(route_key) or []
|
||||
if not only_mcp:
|
||||
continue
|
||||
if not include_connector_route(route_key, available_connectors):
|
||||
continue
|
||||
desc = _MCP_ONLY_ROUTE_DESCRIPTIONS.get(
|
||||
route_key,
|
||||
f"Use for {route_key} tasks related to that system's core work objects and workflows.",
|
||||
)
|
||||
specs.append(
|
||||
DomainRoutingSpec(
|
||||
tool_name=route_key,
|
||||
description=desc,
|
||||
domain_agent=build_mcp_route_domain_agent(llm, route_key, only_mcp),
|
||||
),
|
||||
)
|
||||
|
||||
return routing_tools_from_specs(specs)
|
||||
|
|
@ -1,5 +0,0 @@
|
|||
"""Supervisor agent graph only; supply routing ``tools`` from ``build_supervisor_routing_tools``."""
|
||||
|
||||
from app.agents.multi_agent_chat.supervisor.graph import build_supervisor_agent
|
||||
|
||||
__all__ = ["build_supervisor_agent"]
|
||||
|
|
@ -1,51 +0,0 @@
|
|||
"""Compile the supervisor agent graph (LangChain ``create_agent`` + caller routing tools)."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from collections.abc import Sequence
|
||||
from typing import Any
|
||||
|
||||
from langchain.agents import create_agent
|
||||
from langchain_core.language_models import BaseChatModel
|
||||
from langchain_core.tools import BaseTool
|
||||
from langgraph.types import Checkpointer
|
||||
|
||||
from app.agents.multi_agent_chat.supervisor.prompt_assembly import (
|
||||
build_supervisor_system_prompt,
|
||||
)
|
||||
|
||||
|
||||
def build_supervisor_agent(
|
||||
llm: BaseChatModel,
|
||||
*,
|
||||
tools: Sequence[BaseTool],
|
||||
checkpointer: Checkpointer | None = None,
|
||||
thread_visibility: Any | None = None,
|
||||
middleware: Sequence[Any] | None = None,
|
||||
context_schema: Any | None = None,
|
||||
citations_enabled: bool = True,
|
||||
):
|
||||
"""Compile the supervisor **agent** (graph). ``tools`` = output of ``build_supervisor_routing_tools``."""
|
||||
system_prompt = build_supervisor_system_prompt(
|
||||
tools,
|
||||
thread_visibility=thread_visibility,
|
||||
citations_enabled=citations_enabled,
|
||||
)
|
||||
kwargs: dict[str, Any] = {
|
||||
"system_prompt": system_prompt,
|
||||
"tools": list(tools),
|
||||
"checkpointer": checkpointer,
|
||||
}
|
||||
if middleware is not None:
|
||||
kwargs["middleware"] = list(middleware)
|
||||
if context_schema is not None:
|
||||
kwargs["context_schema"] = context_schema
|
||||
agent = create_agent(llm, **kwargs)
|
||||
if middleware is not None or context_schema is not None:
|
||||
return agent.with_config(
|
||||
{
|
||||
"recursion_limit": 10_000,
|
||||
"metadata": {"ls_integration": "multi_agent_supervisor"},
|
||||
}
|
||||
)
|
||||
return agent
|
||||
|
|
@ -1,128 +0,0 @@
|
|||
"""Supervisor system prompt: template load, shared agent-identity injection, specialist list."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from collections.abc import Sequence
|
||||
from datetime import UTC, datetime
|
||||
from typing import Any
|
||||
|
||||
from langchain_core.tools import BaseTool
|
||||
|
||||
import app.agents.multi_agent_chat.supervisor as supervisor_pkg
|
||||
from app.agents.multi_agent_chat.core.prompts import read_prompt_md
|
||||
from app.agents.new_chat.prompts.composer import _build_citation_block, _read_fragment
|
||||
from app.db import ChatVisibility
|
||||
|
||||
_MEMORY_SPECIALIST_PHRASE = "invoke the **memory** specialist"
|
||||
|
||||
_BUILTIN_SPECIALISTS: frozenset[str] = frozenset({"research", "memory", "deliverables"})
|
||||
_SPECIALIST_CAPABILITIES: dict[str, str] = {
|
||||
"research": "external research: web lookup, source gathering, and SurfSense documentation help.",
|
||||
"memory": "save durable long-lived memory items.",
|
||||
"deliverables": "deliverables and shareable artifacts: reports, podcasts, video presentations, resumes, and images.",
|
||||
"gmail": "email inbox actions: search/read emails, draft updates, send messages, and trash emails.",
|
||||
"calendar": "scheduling actions: check availability, inspect events, create events, and update events.",
|
||||
"google_drive": "Drive file/document actions: locate files, inspect content, and manage files/folders.",
|
||||
"notion": "Notion page actions: create pages, update content, and delete pages.",
|
||||
"confluence": "Confluence page actions: find/read pages and create/update pages.",
|
||||
"dropbox": "Dropbox file storage actions: browse folders, read files, and manage file content.",
|
||||
"onedrive": "OneDrive file storage actions: browse folders, read files, and manage file content.",
|
||||
"discord": "Discord communication actions: read channels/threads and post replies.",
|
||||
"teams": "Microsoft Teams communication actions: read channels/threads and post replies.",
|
||||
"luma": "Luma event actions: list events, inspect event details, and create events.",
|
||||
"linear": "Linear workflow actions: search/update issues and inspect projects/cycles.",
|
||||
"jira": "Jira workflow actions: search/update issues and manage workflow transitions.",
|
||||
"clickup": "ClickUp workflow actions: find/update tasks and lists.",
|
||||
"airtable": "Airtable data actions: locate bases/tables and create/read/update records.",
|
||||
"slack": "Slack communication actions: read channel/thread history and post replies.",
|
||||
# generic_mcp specialist intentionally disabled for now.
|
||||
# "generic_mcp": "handle tasks through user-defined custom app integration tools not covered above.",
|
||||
}
|
||||
_SPECIALIST_ORDER: tuple[str, ...] = tuple(_SPECIALIST_CAPABILITIES.keys())
|
||||
|
||||
|
||||
def _normalize_chat_visibility(thread_visibility: Any | None) -> ChatVisibility:
|
||||
if thread_visibility is None:
|
||||
return ChatVisibility.PRIVATE
|
||||
if thread_visibility == ChatVisibility.SEARCH_SPACE:
|
||||
return ChatVisibility.SEARCH_SPACE
|
||||
raw = getattr(thread_visibility, "value", thread_visibility)
|
||||
if str(raw).upper() == "SEARCH_SPACE":
|
||||
return ChatVisibility.SEARCH_SPACE
|
||||
return ChatVisibility.PRIVATE
|
||||
|
||||
|
||||
def _identity_fragment_key(thread_visibility: Any | None) -> str:
|
||||
"""``private`` / ``team`` suffix for ``agent_*`` and ``memory_protocol_*`` fragments."""
|
||||
return (
|
||||
"team"
|
||||
if _normalize_chat_visibility(thread_visibility) == ChatVisibility.SEARCH_SPACE
|
||||
else "private"
|
||||
)
|
||||
|
||||
|
||||
def _compose_identity_memory_citations(
|
||||
*,
|
||||
thread_visibility: Any | None,
|
||||
citations_enabled: bool,
|
||||
) -> str:
|
||||
"""Main-chat identity, memory protocol, and citation fragments (supervisor slice only)."""
|
||||
key = _identity_fragment_key(thread_visibility)
|
||||
today = datetime.now(UTC).date().isoformat()
|
||||
|
||||
intro = _read_fragment(f"base/agent_{key}.md")
|
||||
if intro:
|
||||
intro = intro.format(resolved_today=today)
|
||||
|
||||
memory = _read_fragment(f"base/memory_protocol_{key}.md").replace(
|
||||
"call update_memory",
|
||||
_MEMORY_SPECIALIST_PHRASE,
|
||||
)
|
||||
tail = (
|
||||
f"<system_instruction>\n{memory}\n\n</system_instruction>\n"
|
||||
+ _build_citation_block(citations_enabled)
|
||||
)
|
||||
return "\n\n".join(part for part in (intro.strip(), tail.strip()) if part)
|
||||
|
||||
|
||||
def _memory_specialist_capability(thread_visibility: Any | None) -> str:
|
||||
vis = str(getattr(thread_visibility, "value", thread_visibility)).upper()
|
||||
if vis == "SEARCH_SPACE":
|
||||
return "team memory actions: save shared team preferences, conventions, and long-lived team facts."
|
||||
return "user memory actions: save personal preferences, instructions, and long-lived user facts."
|
||||
|
||||
|
||||
def _specialists_markdown(
|
||||
tools: Sequence[BaseTool],
|
||||
*,
|
||||
thread_visibility: Any | None,
|
||||
) -> str:
|
||||
available_names = {
|
||||
tool.name for tool in tools if isinstance(getattr(tool, "name", None), str)
|
||||
}
|
||||
capabilities = dict(_SPECIALIST_CAPABILITIES)
|
||||
capabilities["memory"] = _memory_specialist_capability(thread_visibility)
|
||||
lines: list[str] = []
|
||||
for name in _SPECIALIST_ORDER:
|
||||
if name in _BUILTIN_SPECIALISTS or name in available_names:
|
||||
lines.append(f"- {name}: {capabilities[name]}")
|
||||
return "\n".join(lines)
|
||||
|
||||
|
||||
def build_supervisor_system_prompt(
|
||||
tools: Sequence[BaseTool],
|
||||
*,
|
||||
thread_visibility: Any | None,
|
||||
citations_enabled: bool,
|
||||
) -> str:
|
||||
"""Load ``supervisor_prompt.md`` and fill placeholders."""
|
||||
template = read_prompt_md(supervisor_pkg.__name__, "supervisor_prompt")
|
||||
specialists = _specialists_markdown(tools, thread_visibility=thread_visibility)
|
||||
injected = _compose_identity_memory_citations(
|
||||
thread_visibility=thread_visibility,
|
||||
citations_enabled=citations_enabled,
|
||||
)
|
||||
return template.replace("{{AVAILABLE_SPECIALISTS_LIST}}", specialists).replace(
|
||||
"{{SUPERVISOR_BASE_INJECTION}}",
|
||||
injected,
|
||||
)
|
||||
|
|
@ -1,67 +0,0 @@
|
|||
{{SUPERVISOR_BASE_INJECTION}}
|
||||
|
||||
<supervisor_role>
|
||||
In this **multi-agent** session you also **coordinate specialists** (listed below): call a specialist only when their domain matches the need; give each call a compact, outcome-focused task; merge structured results into one clear user-facing reply. When you can satisfy the turn with your own tools and reasoning, do so without delegating.
|
||||
</supervisor_role>
|
||||
|
||||
<available_specialists>
|
||||
Use only the specialists listed below.
|
||||
{{AVAILABLE_SPECIALISTS_LIST}}
|
||||
</available_specialists>
|
||||
|
||||
<delegation_policy>
|
||||
1) Delegate when the request clearly belongs to a specialist's capabilities.
|
||||
2) Answer directly when no expert tool is needed.
|
||||
3) For multi-domain work, decompose into sequential expert calls (or parallel only when independent).
|
||||
4) Do not call a specialist "just in case". Every delegation must have a clear purpose.
|
||||
5) Specialists are best for **one clear step at a time**—for example “find this,” “show that record,” “make this one change.” Do **not** hand them an entire “analyze everything and write me a trends report” brief in one go.
|
||||
6) When the user wants **big-picture synthesis**—patterns across lots of items, comparisons across time, or an executive-style overview—**you** split the work: several **small** asks to whoever actually holds that information (each with a clear cap: how many items, how far back, which fields), then **you** combine the answers into one clear reply. If they need a **deliverable**—a real **artifact** others can read, hear, or watch (report, slide-style video, podcast, resume, image)—delegate to the **deliverables** specialist. Do not ask other specialists to replace that: their job is smaller steps (lookups and targeted changes), not producing the final artifact.
|
||||
7) Each specialist answers in a **single short structured reply** (no extra chatter after it). Ask them only for what that reply can reasonably hold. If the user needs a long narrative or full report, **you** combine steps—or use the **deliverables** specialist—not one overloaded ask.
|
||||
8) Prefer **a few clear, small asks** over one huge vague ask that invites guessing, cut-off answers, or broken replies.
|
||||
</delegation_policy>
|
||||
|
||||
<task_writing_policy>
|
||||
When delegating to a specialist, pass a compact but complete task that includes:
|
||||
- the **outcome** they should produce, in **your own words** as clear instructions (do **not** paste or forward the user’s message verbatim),
|
||||
- concrete limits (dates, names, “last N items,” which details matter),
|
||||
- how you will judge success,
|
||||
- any identifiers or links the user already gave.
|
||||
|
||||
When asking for lists or searches, always say **how many** items at most and **which details** you need back.
|
||||
|
||||
Never pass implementation chatter. Pass only actionable instructions.
|
||||
Each delegation should sound like **one clear action** (or two that belong together), not a full project brief—unless you are intentionally speaking to **research** or to **deliverables** for a **deliverable artifact** (report, slide-style video, podcast, resume, image).
|
||||
</task_writing_policy>
|
||||
|
||||
<expert_output_contract_policy>
|
||||
Every specialist returns **one structured reply** in a fixed layout. Treat it like a small form, not prose. It includes:
|
||||
- **outcome**: succeeded, partly done, blocked, or failed
|
||||
- **short summary** of what they did
|
||||
- **proof**: what they actually saw or changed (when relevant)
|
||||
- **what to do next** if they are not done
|
||||
- **what you must ask the user** if something was missing
|
||||
- **what they assumed** if they had to fill a gap
|
||||
|
||||
How to use it:
|
||||
1) **Succeeded**: only treat it as done if the **proof** backs it up.
|
||||
2) **Partly done**: use what they proved, then follow their **what to do next**.
|
||||
3) **Blocked**: do not blindly retry; ask the user only what they said was missing (or pick from options they listed).
|
||||
4) **Failed**: do not pretend it worked; either retry with a clearer small ask or explain honestly and follow their suggested recovery.
|
||||
5) If the reply is missing, garbled, or contradicts itself, treat it as failed, do not invent facts, and recover with a safer smaller ask or a question to the user.
|
||||
</expert_output_contract_policy>
|
||||
|
||||
<clarification_policy>
|
||||
Ask a concise clarifying question only when a missing detail blocks execution.
|
||||
If one reasonable default is safe and obvious, use it and state the assumption.
|
||||
</clarification_policy>
|
||||
|
||||
<synthesis_policy>
|
||||
After expert calls, produce one coherent final answer:
|
||||
- what was done,
|
||||
- key results/artifacts,
|
||||
- unresolved items and the next best step.
|
||||
- include assumptions only when they affected outcomes.
|
||||
- when multiple experts are used, merge outputs into one user-facing narrative (do not paste their raw structured reply verbatim).
|
||||
|
||||
Never claim an action succeeded unless their reply includes proof that matches what you claim.
|
||||
</synthesis_policy>
|
||||
Loading…
Add table
Add a link
Reference in a new issue