2024-11-07 21:01:51 +00:00
|
|
|
"""
|
|
|
|
|
Invokes the text completion service by specifying an LLM system prompt
|
|
|
|
|
and user prompt. Both arguments are required.
|
|
|
|
|
"""
|
|
|
|
|
|
|
|
|
|
import argparse
|
|
|
|
|
import os
|
2025-12-04 17:38:57 +00:00
|
|
|
from trustgraph.api import Api
|
2024-11-07 21:01:51 +00:00
|
|
|
|
2025-12-04 17:38:57 +00:00
|
|
|
default_url = os.getenv("TRUSTGRAPH_URL", 'http://localhost:8088/')
|
|
|
|
|
default_token = os.getenv("TRUSTGRAPH_TOKEN", None)
|
feat: workspace-based multi-tenancy, replacing user as tenancy axis (#840)
Introduces `workspace` as the isolation boundary for config, flows,
library, and knowledge data. Removes `user` as a schema-level field
throughout the code, API specs, and tests; workspace provides the
same separation more cleanly at the trusted flow.workspace layer
rather than through client-supplied message fields.
Design
------
- IAM tech spec (docs/tech-specs/iam.md) documents current state,
proposed auth/access model, and migration direction.
- Data ownership model (docs/tech-specs/data-ownership-model.md)
captures the workspace/collection/flow hierarchy.
Schema + messaging
------------------
- Drop `user` field from AgentRequest/Step, GraphRagQuery,
DocumentRagQuery, Triples/Graph/Document/Row EmbeddingsRequest,
Sparql/Rows/Structured QueryRequest, ToolServiceRequest.
- Keep collection/workspace routing via flow.workspace at the
service layer.
- Translators updated to not serialise/deserialise user.
API specs
---------
- OpenAPI schemas and path examples cleaned of user fields.
- Websocket async-api messages updated.
- Removed the unused parameters/User.yaml.
Services + base
---------------
- Librarian, collection manager, knowledge, config: all operations
scoped by workspace. Config client API takes workspace as first
positional arg.
- `flow.workspace` set at flow start time by the infrastructure;
no longer pass-through from clients.
- Tool service drops user-personalisation passthrough.
CLI + SDK
---------
- tg-init-workspace and workspace-aware import/export.
- All tg-* commands drop user args; accept --workspace.
- Python API/SDK (flow, socket_client, async_*, explainability,
library) drop user kwargs from every method signature.
MCP server
----------
- All tool endpoints drop user parameters; socket_manager no longer
keyed per user.
Flow service
------------
- Closure-based topic cleanup on flow stop: only delete topics
whose blueprint template was parameterised AND no remaining
live flow (across all workspaces) still resolves to that topic.
Three scopes fall out naturally from template analysis:
* {id} -> per-flow, deleted on stop
* {blueprint} -> per-blueprint, kept while any flow of the
same blueprint exists
* {workspace} -> per-workspace, kept while any flow in the
workspace exists
* literal -> global, never deleted (e.g. tg.request.librarian)
Fixes a bug where stopping a flow silently destroyed the global
librarian exchange, wedging all library operations until manual
restart.
RabbitMQ backend
----------------
- heartbeat=60, blocked_connection_timeout=300. Catches silently
dead connections (broker restart, orphaned channels, network
partitions) within ~2 heartbeat windows, so the consumer
reconnects and re-binds its queue rather than sitting forever
on a zombie connection.
Tests
-----
- Full test refresh: unit, integration, contract, provenance.
- Dropped user-field assertions and constructor kwargs across
~100 test files.
- Renamed user-collection isolation tests to workspace-collection.
2026-04-21 23:23:01 +01:00
|
|
|
default_workspace = os.getenv("TRUSTGRAPH_WORKSPACE", "default")
|
2024-11-07 21:01:51 +00:00
|
|
|
|
Expose LLM token usage across all service layers (#782)
Expose LLM token usage (in_token, out_token, model) across all
service layers
Propagate token counts from LLM services through the prompt,
text-completion, graph-RAG, document-RAG, and agent orchestrator
pipelines to the API gateway and Python SDK. All fields are Optional
— None means "not available", distinguishing from a real zero count.
Key changes:
- Schema: Add in_token/out_token/model to TextCompletionResponse,
PromptResponse, GraphRagResponse, DocumentRagResponse,
AgentResponse
- TextCompletionClient: New TextCompletionResult return type. Split
into text_completion() (non-streaming) and
text_completion_stream() (streaming with per-chunk handler
callback)
- PromptClient: New PromptResult with response_type
(text/json/jsonl), typed fields (text/object/objects), and token
usage. All callers updated.
- RAG services: Accumulate token usage across all prompt calls
(extract-concepts, edge-scoring, edge-reasoning,
synthesis). Non-streaming path sends single combined response
instead of chunk + end_of_session.
- Agent orchestrator: UsageTracker accumulates tokens across
meta-router, pattern prompt calls, and react reasoning. Attached
to end_of_dialog.
- Translators: Encode token fields when not None (is not None, not truthy)
- Python SDK: RAG and text-completion methods return
TextCompletionResult (non-streaming) or RAGChunk/AgentAnswer with
token fields (streaming)
- CLI: --show-usage flag on tg-invoke-llm, tg-invoke-prompt,
tg-invoke-graph-rag, tg-invoke-document-rag, tg-invoke-agent
2026-04-13 14:38:34 +01:00
|
|
|
def query(url, flow_id, system, prompt, streaming=True, token=None,
|
feat: workspace-based multi-tenancy, replacing user as tenancy axis (#840)
Introduces `workspace` as the isolation boundary for config, flows,
library, and knowledge data. Removes `user` as a schema-level field
throughout the code, API specs, and tests; workspace provides the
same separation more cleanly at the trusted flow.workspace layer
rather than through client-supplied message fields.
Design
------
- IAM tech spec (docs/tech-specs/iam.md) documents current state,
proposed auth/access model, and migration direction.
- Data ownership model (docs/tech-specs/data-ownership-model.md)
captures the workspace/collection/flow hierarchy.
Schema + messaging
------------------
- Drop `user` field from AgentRequest/Step, GraphRagQuery,
DocumentRagQuery, Triples/Graph/Document/Row EmbeddingsRequest,
Sparql/Rows/Structured QueryRequest, ToolServiceRequest.
- Keep collection/workspace routing via flow.workspace at the
service layer.
- Translators updated to not serialise/deserialise user.
API specs
---------
- OpenAPI schemas and path examples cleaned of user fields.
- Websocket async-api messages updated.
- Removed the unused parameters/User.yaml.
Services + base
---------------
- Librarian, collection manager, knowledge, config: all operations
scoped by workspace. Config client API takes workspace as first
positional arg.
- `flow.workspace` set at flow start time by the infrastructure;
no longer pass-through from clients.
- Tool service drops user-personalisation passthrough.
CLI + SDK
---------
- tg-init-workspace and workspace-aware import/export.
- All tg-* commands drop user args; accept --workspace.
- Python API/SDK (flow, socket_client, async_*, explainability,
library) drop user kwargs from every method signature.
MCP server
----------
- All tool endpoints drop user parameters; socket_manager no longer
keyed per user.
Flow service
------------
- Closure-based topic cleanup on flow stop: only delete topics
whose blueprint template was parameterised AND no remaining
live flow (across all workspaces) still resolves to that topic.
Three scopes fall out naturally from template analysis:
* {id} -> per-flow, deleted on stop
* {blueprint} -> per-blueprint, kept while any flow of the
same blueprint exists
* {workspace} -> per-workspace, kept while any flow in the
workspace exists
* literal -> global, never deleted (e.g. tg.request.librarian)
Fixes a bug where stopping a flow silently destroyed the global
librarian exchange, wedging all library operations until manual
restart.
RabbitMQ backend
----------------
- heartbeat=60, blocked_connection_timeout=300. Catches silently
dead connections (broker restart, orphaned channels, network
partitions) within ~2 heartbeat windows, so the consumer
reconnects and re-binds its queue rather than sitting forever
on a zombie connection.
Tests
-----
- Full test refresh: unit, integration, contract, provenance.
- Dropped user-field assertions and constructor kwargs across
~100 test files.
- Renamed user-collection isolation tests to workspace-collection.
2026-04-21 23:23:01 +01:00
|
|
|
show_usage=False, workspace="default"):
|
2024-11-07 21:01:51 +00:00
|
|
|
|
2025-12-04 17:38:57 +00:00
|
|
|
# Create API client
|
feat: workspace-based multi-tenancy, replacing user as tenancy axis (#840)
Introduces `workspace` as the isolation boundary for config, flows,
library, and knowledge data. Removes `user` as a schema-level field
throughout the code, API specs, and tests; workspace provides the
same separation more cleanly at the trusted flow.workspace layer
rather than through client-supplied message fields.
Design
------
- IAM tech spec (docs/tech-specs/iam.md) documents current state,
proposed auth/access model, and migration direction.
- Data ownership model (docs/tech-specs/data-ownership-model.md)
captures the workspace/collection/flow hierarchy.
Schema + messaging
------------------
- Drop `user` field from AgentRequest/Step, GraphRagQuery,
DocumentRagQuery, Triples/Graph/Document/Row EmbeddingsRequest,
Sparql/Rows/Structured QueryRequest, ToolServiceRequest.
- Keep collection/workspace routing via flow.workspace at the
service layer.
- Translators updated to not serialise/deserialise user.
API specs
---------
- OpenAPI schemas and path examples cleaned of user fields.
- Websocket async-api messages updated.
- Removed the unused parameters/User.yaml.
Services + base
---------------
- Librarian, collection manager, knowledge, config: all operations
scoped by workspace. Config client API takes workspace as first
positional arg.
- `flow.workspace` set at flow start time by the infrastructure;
no longer pass-through from clients.
- Tool service drops user-personalisation passthrough.
CLI + SDK
---------
- tg-init-workspace and workspace-aware import/export.
- All tg-* commands drop user args; accept --workspace.
- Python API/SDK (flow, socket_client, async_*, explainability,
library) drop user kwargs from every method signature.
MCP server
----------
- All tool endpoints drop user parameters; socket_manager no longer
keyed per user.
Flow service
------------
- Closure-based topic cleanup on flow stop: only delete topics
whose blueprint template was parameterised AND no remaining
live flow (across all workspaces) still resolves to that topic.
Three scopes fall out naturally from template analysis:
* {id} -> per-flow, deleted on stop
* {blueprint} -> per-blueprint, kept while any flow of the
same blueprint exists
* {workspace} -> per-workspace, kept while any flow in the
workspace exists
* literal -> global, never deleted (e.g. tg.request.librarian)
Fixes a bug where stopping a flow silently destroyed the global
librarian exchange, wedging all library operations until manual
restart.
RabbitMQ backend
----------------
- heartbeat=60, blocked_connection_timeout=300. Catches silently
dead connections (broker restart, orphaned channels, network
partitions) within ~2 heartbeat windows, so the consumer
reconnects and re-binds its queue rather than sitting forever
on a zombie connection.
Tests
-----
- Full test refresh: unit, integration, contract, provenance.
- Dropped user-field assertions and constructor kwargs across
~100 test files.
- Renamed user-collection isolation tests to workspace-collection.
2026-04-21 23:23:01 +01:00
|
|
|
api = Api(url=url, token=token, workspace=workspace)
|
2025-12-04 17:38:57 +00:00
|
|
|
socket = api.socket()
|
|
|
|
|
flow = socket.flow(flow_id)
|
2024-11-07 21:01:51 +00:00
|
|
|
|
2025-12-04 17:38:57 +00:00
|
|
|
try:
|
|
|
|
|
# Call text completion
|
|
|
|
|
response = flow.text_completion(
|
|
|
|
|
system=system,
|
|
|
|
|
prompt=prompt,
|
|
|
|
|
streaming=streaming
|
|
|
|
|
)
|
|
|
|
|
|
|
|
|
|
if streaming:
|
Expose LLM token usage across all service layers (#782)
Expose LLM token usage (in_token, out_token, model) across all
service layers
Propagate token counts from LLM services through the prompt,
text-completion, graph-RAG, document-RAG, and agent orchestrator
pipelines to the API gateway and Python SDK. All fields are Optional
— None means "not available", distinguishing from a real zero count.
Key changes:
- Schema: Add in_token/out_token/model to TextCompletionResponse,
PromptResponse, GraphRagResponse, DocumentRagResponse,
AgentResponse
- TextCompletionClient: New TextCompletionResult return type. Split
into text_completion() (non-streaming) and
text_completion_stream() (streaming with per-chunk handler
callback)
- PromptClient: New PromptResult with response_type
(text/json/jsonl), typed fields (text/object/objects), and token
usage. All callers updated.
- RAG services: Accumulate token usage across all prompt calls
(extract-concepts, edge-scoring, edge-reasoning,
synthesis). Non-streaming path sends single combined response
instead of chunk + end_of_session.
- Agent orchestrator: UsageTracker accumulates tokens across
meta-router, pattern prompt calls, and react reasoning. Attached
to end_of_dialog.
- Translators: Encode token fields when not None (is not None, not truthy)
- Python SDK: RAG and text-completion methods return
TextCompletionResult (non-streaming) or RAGChunk/AgentAnswer with
token fields (streaming)
- CLI: --show-usage flag on tg-invoke-llm, tg-invoke-prompt,
tg-invoke-graph-rag, tg-invoke-document-rag, tg-invoke-agent
2026-04-13 14:38:34 +01:00
|
|
|
last_chunk = None
|
2025-12-04 17:38:57 +00:00
|
|
|
for chunk in response:
|
Expose LLM token usage across all service layers (#782)
Expose LLM token usage (in_token, out_token, model) across all
service layers
Propagate token counts from LLM services through the prompt,
text-completion, graph-RAG, document-RAG, and agent orchestrator
pipelines to the API gateway and Python SDK. All fields are Optional
— None means "not available", distinguishing from a real zero count.
Key changes:
- Schema: Add in_token/out_token/model to TextCompletionResponse,
PromptResponse, GraphRagResponse, DocumentRagResponse,
AgentResponse
- TextCompletionClient: New TextCompletionResult return type. Split
into text_completion() (non-streaming) and
text_completion_stream() (streaming with per-chunk handler
callback)
- PromptClient: New PromptResult with response_type
(text/json/jsonl), typed fields (text/object/objects), and token
usage. All callers updated.
- RAG services: Accumulate token usage across all prompt calls
(extract-concepts, edge-scoring, edge-reasoning,
synthesis). Non-streaming path sends single combined response
instead of chunk + end_of_session.
- Agent orchestrator: UsageTracker accumulates tokens across
meta-router, pattern prompt calls, and react reasoning. Attached
to end_of_dialog.
- Translators: Encode token fields when not None (is not None, not truthy)
- Python SDK: RAG and text-completion methods return
TextCompletionResult (non-streaming) or RAGChunk/AgentAnswer with
token fields (streaming)
- CLI: --show-usage flag on tg-invoke-llm, tg-invoke-prompt,
tg-invoke-graph-rag, tg-invoke-document-rag, tg-invoke-agent
2026-04-13 14:38:34 +01:00
|
|
|
print(chunk.content, end="", flush=True)
|
|
|
|
|
last_chunk = chunk
|
2025-12-04 17:38:57 +00:00
|
|
|
print()
|
Expose LLM token usage across all service layers (#782)
Expose LLM token usage (in_token, out_token, model) across all
service layers
Propagate token counts from LLM services through the prompt,
text-completion, graph-RAG, document-RAG, and agent orchestrator
pipelines to the API gateway and Python SDK. All fields are Optional
— None means "not available", distinguishing from a real zero count.
Key changes:
- Schema: Add in_token/out_token/model to TextCompletionResponse,
PromptResponse, GraphRagResponse, DocumentRagResponse,
AgentResponse
- TextCompletionClient: New TextCompletionResult return type. Split
into text_completion() (non-streaming) and
text_completion_stream() (streaming with per-chunk handler
callback)
- PromptClient: New PromptResult with response_type
(text/json/jsonl), typed fields (text/object/objects), and token
usage. All callers updated.
- RAG services: Accumulate token usage across all prompt calls
(extract-concepts, edge-scoring, edge-reasoning,
synthesis). Non-streaming path sends single combined response
instead of chunk + end_of_session.
- Agent orchestrator: UsageTracker accumulates tokens across
meta-router, pattern prompt calls, and react reasoning. Attached
to end_of_dialog.
- Translators: Encode token fields when not None (is not None, not truthy)
- Python SDK: RAG and text-completion methods return
TextCompletionResult (non-streaming) or RAGChunk/AgentAnswer with
token fields (streaming)
- CLI: --show-usage flag on tg-invoke-llm, tg-invoke-prompt,
tg-invoke-graph-rag, tg-invoke-document-rag, tg-invoke-agent
2026-04-13 14:38:34 +01:00
|
|
|
|
|
|
|
|
if show_usage and last_chunk:
|
|
|
|
|
print(
|
|
|
|
|
f"Input tokens: {last_chunk.in_token} "
|
|
|
|
|
f"Output tokens: {last_chunk.out_token} "
|
|
|
|
|
f"Model: {last_chunk.model}",
|
|
|
|
|
file=__import__('sys').stderr,
|
|
|
|
|
)
|
2025-12-04 17:38:57 +00:00
|
|
|
else:
|
Expose LLM token usage across all service layers (#782)
Expose LLM token usage (in_token, out_token, model) across all
service layers
Propagate token counts from LLM services through the prompt,
text-completion, graph-RAG, document-RAG, and agent orchestrator
pipelines to the API gateway and Python SDK. All fields are Optional
— None means "not available", distinguishing from a real zero count.
Key changes:
- Schema: Add in_token/out_token/model to TextCompletionResponse,
PromptResponse, GraphRagResponse, DocumentRagResponse,
AgentResponse
- TextCompletionClient: New TextCompletionResult return type. Split
into text_completion() (non-streaming) and
text_completion_stream() (streaming with per-chunk handler
callback)
- PromptClient: New PromptResult with response_type
(text/json/jsonl), typed fields (text/object/objects), and token
usage. All callers updated.
- RAG services: Accumulate token usage across all prompt calls
(extract-concepts, edge-scoring, edge-reasoning,
synthesis). Non-streaming path sends single combined response
instead of chunk + end_of_session.
- Agent orchestrator: UsageTracker accumulates tokens across
meta-router, pattern prompt calls, and react reasoning. Attached
to end_of_dialog.
- Translators: Encode token fields when not None (is not None, not truthy)
- Python SDK: RAG and text-completion methods return
TextCompletionResult (non-streaming) or RAGChunk/AgentAnswer with
token fields (streaming)
- CLI: --show-usage flag on tg-invoke-llm, tg-invoke-prompt,
tg-invoke-graph-rag, tg-invoke-document-rag, tg-invoke-agent
2026-04-13 14:38:34 +01:00
|
|
|
print(response.text)
|
|
|
|
|
|
|
|
|
|
if show_usage:
|
|
|
|
|
print(
|
|
|
|
|
f"Input tokens: {response.in_token} "
|
|
|
|
|
f"Output tokens: {response.out_token} "
|
|
|
|
|
f"Model: {response.model}",
|
|
|
|
|
file=__import__('sys').stderr,
|
|
|
|
|
)
|
2025-12-04 17:38:57 +00:00
|
|
|
|
|
|
|
|
finally:
|
|
|
|
|
# Clean up socket connection
|
|
|
|
|
socket.close()
|
2024-11-07 21:01:51 +00:00
|
|
|
|
|
|
|
|
def main():
|
|
|
|
|
|
|
|
|
|
parser = argparse.ArgumentParser(
|
|
|
|
|
prog='tg-invoke-llm',
|
|
|
|
|
description=__doc__,
|
|
|
|
|
)
|
|
|
|
|
|
|
|
|
|
parser.add_argument(
|
2025-01-02 19:49:22 +00:00
|
|
|
'-u', '--url',
|
|
|
|
|
default=default_url,
|
|
|
|
|
help=f'API URL (default: {default_url})',
|
2024-11-07 21:01:51 +00:00
|
|
|
)
|
|
|
|
|
|
2025-12-04 17:38:57 +00:00
|
|
|
parser.add_argument(
|
|
|
|
|
'-t', '--token',
|
|
|
|
|
default=default_token,
|
|
|
|
|
help='Authentication token (default: $TRUSTGRAPH_TOKEN)',
|
|
|
|
|
)
|
|
|
|
|
|
feat: workspace-based multi-tenancy, replacing user as tenancy axis (#840)
Introduces `workspace` as the isolation boundary for config, flows,
library, and knowledge data. Removes `user` as a schema-level field
throughout the code, API specs, and tests; workspace provides the
same separation more cleanly at the trusted flow.workspace layer
rather than through client-supplied message fields.
Design
------
- IAM tech spec (docs/tech-specs/iam.md) documents current state,
proposed auth/access model, and migration direction.
- Data ownership model (docs/tech-specs/data-ownership-model.md)
captures the workspace/collection/flow hierarchy.
Schema + messaging
------------------
- Drop `user` field from AgentRequest/Step, GraphRagQuery,
DocumentRagQuery, Triples/Graph/Document/Row EmbeddingsRequest,
Sparql/Rows/Structured QueryRequest, ToolServiceRequest.
- Keep collection/workspace routing via flow.workspace at the
service layer.
- Translators updated to not serialise/deserialise user.
API specs
---------
- OpenAPI schemas and path examples cleaned of user fields.
- Websocket async-api messages updated.
- Removed the unused parameters/User.yaml.
Services + base
---------------
- Librarian, collection manager, knowledge, config: all operations
scoped by workspace. Config client API takes workspace as first
positional arg.
- `flow.workspace` set at flow start time by the infrastructure;
no longer pass-through from clients.
- Tool service drops user-personalisation passthrough.
CLI + SDK
---------
- tg-init-workspace and workspace-aware import/export.
- All tg-* commands drop user args; accept --workspace.
- Python API/SDK (flow, socket_client, async_*, explainability,
library) drop user kwargs from every method signature.
MCP server
----------
- All tool endpoints drop user parameters; socket_manager no longer
keyed per user.
Flow service
------------
- Closure-based topic cleanup on flow stop: only delete topics
whose blueprint template was parameterised AND no remaining
live flow (across all workspaces) still resolves to that topic.
Three scopes fall out naturally from template analysis:
* {id} -> per-flow, deleted on stop
* {blueprint} -> per-blueprint, kept while any flow of the
same blueprint exists
* {workspace} -> per-workspace, kept while any flow in the
workspace exists
* literal -> global, never deleted (e.g. tg.request.librarian)
Fixes a bug where stopping a flow silently destroyed the global
librarian exchange, wedging all library operations until manual
restart.
RabbitMQ backend
----------------
- heartbeat=60, blocked_connection_timeout=300. Catches silently
dead connections (broker restart, orphaned channels, network
partitions) within ~2 heartbeat windows, so the consumer
reconnects and re-binds its queue rather than sitting forever
on a zombie connection.
Tests
-----
- Full test refresh: unit, integration, contract, provenance.
- Dropped user-field assertions and constructor kwargs across
~100 test files.
- Renamed user-collection isolation tests to workspace-collection.
2026-04-21 23:23:01 +01:00
|
|
|
parser.add_argument(
|
|
|
|
|
'-w', '--workspace',
|
|
|
|
|
default=default_workspace,
|
|
|
|
|
help=f'Workspace (default: {default_workspace})',
|
|
|
|
|
)
|
|
|
|
|
|
2024-11-07 21:01:51 +00:00
|
|
|
parser.add_argument(
|
|
|
|
|
'system',
|
|
|
|
|
nargs=1,
|
|
|
|
|
help='LLM system prompt e.g. You are a helpful assistant',
|
|
|
|
|
)
|
|
|
|
|
|
|
|
|
|
parser.add_argument(
|
|
|
|
|
'prompt',
|
|
|
|
|
nargs=1,
|
|
|
|
|
help='LLM prompt e.g. What is 2 + 2?',
|
|
|
|
|
)
|
2025-05-03 10:39:53 +01:00
|
|
|
|
|
|
|
|
parser.add_argument(
|
|
|
|
|
'-f', '--flow-id',
|
2025-05-24 12:27:56 +01:00
|
|
|
default="default",
|
|
|
|
|
help=f'Flow ID (default: default)'
|
2025-05-03 10:39:53 +01:00
|
|
|
)
|
2024-11-07 21:01:51 +00:00
|
|
|
|
2025-11-26 09:59:10 +00:00
|
|
|
parser.add_argument(
|
|
|
|
|
'--no-streaming',
|
|
|
|
|
action='store_true',
|
|
|
|
|
help='Disable streaming (default: streaming enabled)'
|
|
|
|
|
)
|
|
|
|
|
|
Expose LLM token usage across all service layers (#782)
Expose LLM token usage (in_token, out_token, model) across all
service layers
Propagate token counts from LLM services through the prompt,
text-completion, graph-RAG, document-RAG, and agent orchestrator
pipelines to the API gateway and Python SDK. All fields are Optional
— None means "not available", distinguishing from a real zero count.
Key changes:
- Schema: Add in_token/out_token/model to TextCompletionResponse,
PromptResponse, GraphRagResponse, DocumentRagResponse,
AgentResponse
- TextCompletionClient: New TextCompletionResult return type. Split
into text_completion() (non-streaming) and
text_completion_stream() (streaming with per-chunk handler
callback)
- PromptClient: New PromptResult with response_type
(text/json/jsonl), typed fields (text/object/objects), and token
usage. All callers updated.
- RAG services: Accumulate token usage across all prompt calls
(extract-concepts, edge-scoring, edge-reasoning,
synthesis). Non-streaming path sends single combined response
instead of chunk + end_of_session.
- Agent orchestrator: UsageTracker accumulates tokens across
meta-router, pattern prompt calls, and react reasoning. Attached
to end_of_dialog.
- Translators: Encode token fields when not None (is not None, not truthy)
- Python SDK: RAG and text-completion methods return
TextCompletionResult (non-streaming) or RAGChunk/AgentAnswer with
token fields (streaming)
- CLI: --show-usage flag on tg-invoke-llm, tg-invoke-prompt,
tg-invoke-graph-rag, tg-invoke-document-rag, tg-invoke-agent
2026-04-13 14:38:34 +01:00
|
|
|
parser.add_argument(
|
|
|
|
|
'--show-usage',
|
|
|
|
|
action='store_true',
|
|
|
|
|
help='Show token usage and model on stderr'
|
|
|
|
|
)
|
|
|
|
|
|
2024-11-07 21:01:51 +00:00
|
|
|
args = parser.parse_args()
|
|
|
|
|
|
|
|
|
|
try:
|
|
|
|
|
|
2025-12-04 17:38:57 +00:00
|
|
|
query(
|
2025-01-02 19:49:22 +00:00
|
|
|
url=args.url,
|
2025-11-26 09:59:10 +00:00
|
|
|
flow_id=args.flow_id,
|
2024-11-07 21:01:51 +00:00
|
|
|
system=args.system[0],
|
|
|
|
|
prompt=args.prompt[0],
|
2025-12-04 17:38:57 +00:00
|
|
|
streaming=not args.no_streaming,
|
|
|
|
|
token=args.token,
|
Expose LLM token usage across all service layers (#782)
Expose LLM token usage (in_token, out_token, model) across all
service layers
Propagate token counts from LLM services through the prompt,
text-completion, graph-RAG, document-RAG, and agent orchestrator
pipelines to the API gateway and Python SDK. All fields are Optional
— None means "not available", distinguishing from a real zero count.
Key changes:
- Schema: Add in_token/out_token/model to TextCompletionResponse,
PromptResponse, GraphRagResponse, DocumentRagResponse,
AgentResponse
- TextCompletionClient: New TextCompletionResult return type. Split
into text_completion() (non-streaming) and
text_completion_stream() (streaming with per-chunk handler
callback)
- PromptClient: New PromptResult with response_type
(text/json/jsonl), typed fields (text/object/objects), and token
usage. All callers updated.
- RAG services: Accumulate token usage across all prompt calls
(extract-concepts, edge-scoring, edge-reasoning,
synthesis). Non-streaming path sends single combined response
instead of chunk + end_of_session.
- Agent orchestrator: UsageTracker accumulates tokens across
meta-router, pattern prompt calls, and react reasoning. Attached
to end_of_dialog.
- Translators: Encode token fields when not None (is not None, not truthy)
- Python SDK: RAG and text-completion methods return
TextCompletionResult (non-streaming) or RAGChunk/AgentAnswer with
token fields (streaming)
- CLI: --show-usage flag on tg-invoke-llm, tg-invoke-prompt,
tg-invoke-graph-rag, tg-invoke-document-rag, tg-invoke-agent
2026-04-13 14:38:34 +01:00
|
|
|
show_usage=args.show_usage,
|
feat: workspace-based multi-tenancy, replacing user as tenancy axis (#840)
Introduces `workspace` as the isolation boundary for config, flows,
library, and knowledge data. Removes `user` as a schema-level field
throughout the code, API specs, and tests; workspace provides the
same separation more cleanly at the trusted flow.workspace layer
rather than through client-supplied message fields.
Design
------
- IAM tech spec (docs/tech-specs/iam.md) documents current state,
proposed auth/access model, and migration direction.
- Data ownership model (docs/tech-specs/data-ownership-model.md)
captures the workspace/collection/flow hierarchy.
Schema + messaging
------------------
- Drop `user` field from AgentRequest/Step, GraphRagQuery,
DocumentRagQuery, Triples/Graph/Document/Row EmbeddingsRequest,
Sparql/Rows/Structured QueryRequest, ToolServiceRequest.
- Keep collection/workspace routing via flow.workspace at the
service layer.
- Translators updated to not serialise/deserialise user.
API specs
---------
- OpenAPI schemas and path examples cleaned of user fields.
- Websocket async-api messages updated.
- Removed the unused parameters/User.yaml.
Services + base
---------------
- Librarian, collection manager, knowledge, config: all operations
scoped by workspace. Config client API takes workspace as first
positional arg.
- `flow.workspace` set at flow start time by the infrastructure;
no longer pass-through from clients.
- Tool service drops user-personalisation passthrough.
CLI + SDK
---------
- tg-init-workspace and workspace-aware import/export.
- All tg-* commands drop user args; accept --workspace.
- Python API/SDK (flow, socket_client, async_*, explainability,
library) drop user kwargs from every method signature.
MCP server
----------
- All tool endpoints drop user parameters; socket_manager no longer
keyed per user.
Flow service
------------
- Closure-based topic cleanup on flow stop: only delete topics
whose blueprint template was parameterised AND no remaining
live flow (across all workspaces) still resolves to that topic.
Three scopes fall out naturally from template analysis:
* {id} -> per-flow, deleted on stop
* {blueprint} -> per-blueprint, kept while any flow of the
same blueprint exists
* {workspace} -> per-workspace, kept while any flow in the
workspace exists
* literal -> global, never deleted (e.g. tg.request.librarian)
Fixes a bug where stopping a flow silently destroyed the global
librarian exchange, wedging all library operations until manual
restart.
RabbitMQ backend
----------------
- heartbeat=60, blocked_connection_timeout=300. Catches silently
dead connections (broker restart, orphaned channels, network
partitions) within ~2 heartbeat windows, so the consumer
reconnects and re-binds its queue rather than sitting forever
on a zombie connection.
Tests
-----
- Full test refresh: unit, integration, contract, provenance.
- Dropped user-field assertions and constructor kwargs across
~100 test files.
- Renamed user-collection isolation tests to workspace-collection.
2026-04-21 23:23:01 +01:00
|
|
|
workspace=args.workspace,
|
2025-12-04 17:38:57 +00:00
|
|
|
)
|
2024-11-07 21:01:51 +00:00
|
|
|
|
|
|
|
|
except Exception as e:
|
|
|
|
|
|
|
|
|
|
print("Exception:", e, flush=True)
|
|
|
|
|
|
2025-07-23 21:22:08 +01:00
|
|
|
if __name__ == "__main__":
|
2025-12-04 17:38:57 +00:00
|
|
|
main()
|