diff --git a/config/vault.example.yaml b/config/vault.example.yaml
new file mode 100644
index 000000000..0e197d2a8
--- /dev/null
+++ b/config/vault.example.yaml
@@ -0,0 +1,48 @@
+# Usage:
+# 1. Get value.
+# >>> from metagpt.tools.libs.env import get_env
+# >>> access_token = await get_env(key="access_token", app_name="github")
+# >>> print(access_token)
+# YOUR_ACCESS_TOKEN
+#
+# 2. Get description for LLM understanding.
+# >>> from metagpt.tools.libs.env import get_env_description
+# >>> descriptions = await get_env_description
+# >>> for k, desc in descriptions.items():
+# >>>     print(f"{key}:{desc}")
+# await get_env(key="access_token", app_name="github"):Get github access token
+# await get_env(key="access_token", app_name="gitlab"):Get gitlab access token
+# ...
+
+vault:
+  github:
+    values:
+      access_token: "YOUR_ACCESS_TOKEN"
+    descriptions:
+      access_token: "Get github access token"
+  gitlab:
+    values:
+      access_token: "YOUR_ACCESS_TOKEN"
+    descriptions:
+      access_token: "Get gitlab access token"
+  iflytek_tts:
+    values:
+      api_id: "YOUR_APP_ID"
+      api_key: "YOUR_API_KEY"
+      api_secret: "YOUR_API_SECRET"
+    descriptions:
+      api_id: "Get the API ID of IFlyTek Text to Speech"
+      api_key: "Get the API KEY of IFlyTek Text to Speech"
+      api_secret: "Get the API SECRET of IFlyTek Text to Speech"
+  azure_tts:
+    values:
+      subscription_key: "YOUR_SUBSCRIPTION_KEY"
+      region: "YOUR_REGION"
+    descriptions:
+      subscription_key: "Get the subscription key of Azure Text to Speech."
+      region: "Get the region of Azure Text to Speech."
+  default: # All key-value pairs whose app name is an empty string are placed below
+    values:
+      proxy: "YOUR_PROXY"
+    descriptions:
+      proxy: "Get proxy for tools like requests, playwright, selenium, etc."
\ No newline at end of file
diff --git a/examples/di/crawl_webpage.py b/examples/di/crawl_webpage.py
index b8226f4f4..10b230f2b 100644
--- a/examples/di/crawl_webpage.py
+++ b/examples/di/crawl_webpage.py
@@ -6,16 +6,19 @@
 """
 
 from metagpt.roles.di.data_interpreter import DataInterpreter
+from metagpt.tools.libs.browser import Browser as _
+
 
 PAPER_LIST_REQ = """"
 Get data from `paperlist` table in https://papercopilot.com/statistics/iclr-statistics/iclr-2024-statistics/,
-and save it to a csv file. paper title must include `multiagent` or `large language model`. *notice: print key variables*
+and save it to a csv file. paper title must include `multiagent` or `large language model`.
+**Notice: view the page element before writing scraping code**
 """
 
 ECOMMERCE_REQ = """
 Get products data from website https://scrapeme.live/shop/ and save it as a csv file.
-**Notice: Firstly parse the web page encoding and the text HTML structure;
-The first page product name, price, product URL, and image URL must be saved in the csv;**
+The first page product name, price, product URL, and image URL must be saved in the csv.
+**Notice: view the page element before writing scraping code**
 """
 
 NEWS_36KR_REQ = """从36kr创投平台https://pitchhub.36kr.com/financing-flash 所有初创企业融资的信息, **注意: 这是一个中文网站**;
@@ -25,11 +28,12 @@ NEWS_36KR_REQ = """从36kr创投平台https://pitchhub.36kr.com/financing-flash
 3. 反思*快讯的html内容示例*中的规律, 设计正则匹配表达式来获取*`快讯`*的标题、链接、时间;
 4. 筛选最近3天的初创企业融资*`快讯`*, 以list[dict]形式打印前5个。
 5. 将全部结果存在本地csv中
+**Notice: view the page element before writing scraping code**
 """
 
 
 async def main():
-    di = DataInterpreter(tools=["scrape_web_playwright"])
+    di = DataInterpreter(tools=["Browser"])
 
     await di.run(ECOMMERCE_REQ)
 
diff --git a/metagpt/actions/di/execute_nb_code.py b/metagpt/actions/di/execute_nb_code.py
index b4fe949fe..64620d9cc 100644
--- a/metagpt/actions/di/execute_nb_code.py
+++ b/metagpt/actions/di/execute_nb_code.py
@@ -65,7 +65,7 @@ class ExecuteNbCode(Action):
     """execute notebook code block, return result to llm, and display it."""
 
     nb: NotebookNode
-    nb_client: NotebookClient = None
+    nb_client: RealtimeOutputNotebookClient = None
     console: Console
     interaction: str
     timeout: int = 600
@@ -78,11 +78,15 @@ class ExecuteNbCode(Action):
             interaction=("ipython" if self.is_ipython() else "terminal"),
         )
         self.reporter = NotebookReporter()
+        self.set_nb_client()
+
+    def set_nb_client(self):
         self.nb_client = RealtimeOutputNotebookClient(
-            nb,
-            timeout=timeout,
+            self.nb,
+            timeout=self.timeout,
             resources={"metadata": {"path": DEFAULT_WORKSPACE_ROOT}},
             notebook_reporter=self.reporter,
+            coalesce_streams=True,
         )
 
     async def build(self):
@@ -118,7 +122,7 @@ class ExecuteNbCode(Action):
         # sleep 1s to wait for the kernel to be cleaned up completely
         await asyncio.sleep(1)
         await self.build()
-        self.nb_client = NotebookClient(self.nb, timeout=self.timeout)
+        self.set_nb_client()
 
     def add_code_cell(self, code: str):
         self.nb.cells.append(new_code_cell(source=code))
diff --git a/metagpt/actions/write_code.py b/metagpt/actions/write_code.py
index 7f225d469..da25fe621 100644
--- a/metagpt/actions/write_code.py
+++ b/metagpt/actions/write_code.py
@@ -161,7 +161,7 @@ class WriteCode(Action):
                     filename=coding_context.filename, root_path=str(self.repo.src_relative_path)
                 )
             coding_context.code_doc.content = code
-            await reporter.async_report(self.repo.workdir / coding_context.code_doc.root_relative_path, "path")
+            await reporter.async_report(coding_context.code_doc, "document")
         return coding_context
 
     @staticmethod
diff --git a/metagpt/actions/write_code_review.py b/metagpt/actions/write_code_review.py
index 3912095df..ad99de2dd 100644
--- a/metagpt/actions/write_code_review.py
+++ b/metagpt/actions/write_code_review.py
@@ -18,6 +18,7 @@ from metagpt.logs import logger
 from metagpt.schema import CodingContext, Document
 from metagpt.utils.common import CodeParser
 from metagpt.utils.project_repo import ProjectRepo
+from metagpt.utils.report import EditorReporter
 
 PROMPT_TEMPLATE = """
 # System
@@ -131,16 +132,23 @@ class WriteCodeReview(Action):
     input_args: Optional[BaseModel] = Field(default=None, exclude=True)
 
     @retry(wait=wait_random_exponential(min=1, max=60), stop=stop_after_attempt(6))
-    async def write_code_review_and_rewrite(self, context_prompt, cr_prompt, filename):
+    async def write_code_review_and_rewrite(self, context_prompt, cr_prompt, doc):
+        filename = doc.filename
         cr_rsp = await self._aask(context_prompt + cr_prompt)
         result = CodeParser.parse_block("Code Review Result", cr_rsp)
         if "LGTM" in result:
             return result, None
 
         # if LBTM, rewrite code
-        rewrite_prompt = f"{context_prompt}\n{cr_rsp}\n{REWRITE_CODE_TEMPLATE.format(filename=filename)}"
-        code_rsp = await self._aask(rewrite_prompt)
-        code = CodeParser.parse_code(text=code_rsp)
+        async with EditorReporter(enable_llm_stream=True) as reporter:
+            await reporter.async_report(
+                {"type": "code", "filename": filename, "src_path": doc.root_relative_path}, "meta"
+            )
+            rewrite_prompt = f"{context_prompt}\n{cr_rsp}\n{REWRITE_CODE_TEMPLATE.format(filename=filename)}"
+            code_rsp = await self._aask(rewrite_prompt)
+            code = CodeParser.parse_code(text=code_rsp)
+            doc.content = code
+            await reporter.async_report(doc, "document")
         return result, code
 
     async def run(self, *args, **kwargs) -> CodingContext:
@@ -185,7 +193,7 @@ class WriteCodeReview(Action):
                 f"len(self.i_context.code_doc.content)={len2}"
             )
             result, rewrited_code = await self.write_code_review_and_rewrite(
-                context_prompt, cr_prompt, self.i_context.code_doc.filename
+                context_prompt, cr_prompt, self.i_context.code_doc
             )
             if "LBTM" in result:
                 iterative_code = rewrited_code
diff --git a/metagpt/rag/engines/simple.py b/metagpt/rag/engines/simple.py
index 5c5810308..623b3f350 100644
--- a/metagpt/rag/engines/simple.py
+++ b/metagpt/rag/engines/simple.py
@@ -4,6 +4,7 @@ import json
 import os
 from typing import Any, Optional, Union
 
+from fsspec import AbstractFileSystem
 from llama_index.core import SimpleDirectoryReader, VectorStoreIndex
 from llama_index.core.callbacks.base import CallbackManager
 from llama_index.core.embeddings import BaseEmbedding
@@ -83,6 +84,7 @@ class SimpleEngine(RetrieverQueryEngine):
         llm: LLM = None,
         retriever_configs: list[BaseRetrieverConfig] = None,
         ranker_configs: list[BaseRankerConfig] = None,
+        fs: Optional[AbstractFileSystem] = None,
     ) -> "SimpleEngine":
         """From docs.
 
@@ -100,7 +102,7 @@ class SimpleEngine(RetrieverQueryEngine):
         if not input_dir and not input_files:
             raise ValueError("Must provide either `input_dir` or `input_files`.")
 
-        documents = SimpleDirectoryReader(input_dir=input_dir, input_files=input_files).load_data()
+        documents = SimpleDirectoryReader(input_dir=input_dir, input_files=input_files, fs=fs).load_data()
         cls._fix_document_metadata(documents)
 
         index = VectorStoreIndex.from_documents(
diff --git a/metagpt/roles/di/data_analyst.py b/metagpt/roles/di/data_analyst.py
index 0fc95b9d6..fc298ea4c 100644
--- a/metagpt/roles/di/data_analyst.py
+++ b/metagpt/roles/di/data_analyst.py
@@ -20,6 +20,7 @@ from metagpt.strategy.thinking_command import (
 )
 from metagpt.tools.tool_recommend import BM25ToolRecommender
 from metagpt.utils.common import CodeParser
+from metagpt.utils.report import ThoughtReporter
 
 
 class DataAnalyst(DataInterpreter):
@@ -82,8 +83,8 @@ class DataAnalyst(DataInterpreter):
             available_commands=prepare_command_prompt(self.available_commands),
         )
         context = self.llm.format_msg(self.working_memory.get() + [Message(content=prompt, role="user")])
-
-        rsp = await self.llm.aask(context)
+        async with ThoughtReporter():
+            rsp = await self.llm.aask(context)
         self.commands = json.loads(CodeParser.parse_code(block=None, text=rsp))
         self.rc.memory.add(Message(content=rsp, role="assistant"))
 
diff --git a/metagpt/roles/di/data_interpreter.py b/metagpt/roles/di/data_interpreter.py
index e147cbbe3..bdfc0e294 100644
--- a/metagpt/roles/di/data_interpreter.py
+++ b/metagpt/roles/di/data_interpreter.py
@@ -15,6 +15,7 @@ from metagpt.schema import Message, Task, TaskResult
 from metagpt.strategy.task_type import TaskType
 from metagpt.tools.tool_recommend import BM25ToolRecommender, ToolRecommender
 from metagpt.utils.common import CodeParser
+from metagpt.utils.report import ThoughtReporter
 
 REACT_THINK_PROMPT = """
 # User Requirement
@@ -73,7 +74,8 @@ class DataInterpreter(Role):
             return True
 
         prompt = REACT_THINK_PROMPT.format(user_requirement=self.user_requirement, context=context)
-        rsp = await self.llm.aask(prompt)
+        async with ThoughtReporter():
+            rsp = await self.llm.aask(prompt)
         rsp_dict = json.loads(CodeParser.parse_code(text=rsp))
         self.working_memory.add(Message(content=rsp_dict["thoughts"], role="assistant"))
         need_action = rsp_dict["state"]
diff --git a/metagpt/roles/di/team_leader.py b/metagpt/roles/di/team_leader.py
index a1ef11fa6..2fa782ade 100644
--- a/metagpt/roles/di/team_leader.py
+++ b/metagpt/roles/di/team_leader.py
@@ -20,6 +20,7 @@ from metagpt.strategy.thinking_command import (
     run_commands,
 )
 from metagpt.utils.common import CodeParser
+from metagpt.utils.report import ThoughtReporter
 
 
 class TeamLeader(Role):
@@ -69,7 +70,8 @@ class TeamLeader(Role):
         )
         context = self.llm.format_msg(self.get_memories(k=10) + [Message(content=prompt, role="user")])
 
-        rsp = await self.llm.aask(context, system_msgs=[SYSTEM_PROMPT])
+        async with ThoughtReporter():
+            rsp = await self.llm.aask(context, system_msgs=[SYSTEM_PROMPT])
         self.commands = json.loads(CodeParser.parse_code(text=rsp))
         self.rc.memory.add(Message(content=rsp, role="assistant"))
 
diff --git a/metagpt/tools/libs/browser.py b/metagpt/tools/libs/browser.py
index 7fde804fe..8d6daec11 100644
--- a/metagpt/tools/libs/browser.py
+++ b/metagpt/tools/libs/browser.py
@@ -1,9 +1,12 @@
 from __future__ import annotations
+import contextlib
 
 from playwright.async_api import async_playwright
-
+from metagpt.utils.file import MemoryFileSystem
+from uuid import uuid4
 from metagpt.const import DEFAULT_WORKSPACE_ROOT
 from metagpt.tools.tool_registry import register_tool
+from metagpt.utils.parse_html import simplify_html
 from metagpt.utils.report import BrowserReporter
 
 
@@ -35,16 +38,49 @@ class Browser:
         print("Now on page ", url)
         await self._view()
 
-    async def open_new_page(self, url: str):
+    async def open_new_page(self, url: str, timeout: float = 30000):
         """open a new page in the browser and view the page"""
         async with self.reporter as reporter:
             page = await self.browser.new_page()
             await reporter.async_report(url, "url")
-            await page.goto(url)
+            await page.goto(url, timeout=timeout)
             self.pages[url] = page
             await self._set_current_page(page, url)
             await reporter.async_report(page, "page")
 
+    async def view_page_element_to_scrape(self, requirement: str, keep_links: bool = False) -> None:
+        """view the HTML content of current page to understand the structure. When executed, the content will be printed out
+
+        Args:
+            requirement (str): Providing a clear and detailed requirement helps in focusing the inspection on the desired elements.
+            keep_links (bool): Whether to keep the hyperlinks in the HTML content. Set to True if links are required
+        """
+        html = await self.current_page.content()
+        html = simplify_html(html, url=self.current_page.url, keep_links=keep_links)
+        mem_fs = MemoryFileSystem()
+        filename = f"{uuid4().hex}.html"
+        with mem_fs.open(filename, "w") as f:
+            f.write(html)
+
+        # Since RAG is an optional optimization, if it fails, the simplified HTML can be used as a fallback.
+        with contextlib.suppress(Exception):
+
+            from metagpt.rag.engines import SimpleEngine  # avoid circular import
+
+            # TODO make `from_docs` asynchronous
+            engine = SimpleEngine.from_docs(input_files=[filename], fs=mem_fs)
+            nodes = await engine.aretrieve(requirement)
+            html = "\n".join(i.text for i in nodes)
+
+        mem_fs.rm_file(filename)
+        print(html)
+
+    async def get_page_content(self) -> str:
+        """Get the HTML content of current page."""
+        html = await self.current_page.content()
+        html_content = html.strip()
+        return html_content
+
     async def switch_page(self, url: str):
         """switch to an opened page in the browser and view the page"""
         if url in self.pages:
@@ -152,8 +188,8 @@ class Browser:
 
     async def _view(self, keep_len: int = 5000) -> str:
         """simulate human viewing the current page, return the visible text with links"""
-        visible_text_with_links = await self.current_page.evaluate(VIEW_CONTENT_JS)
-        print("The visible text and their links (if any): ", visible_text_with_links[:keep_len])
+        # visible_text_with_links = await self.current_page.evaluate(VIEW_CONTENT_JS)
+        # print("The visible text and their links (if any): ", visible_text_with_links[:keep_len])
         # html_content = await self._view_page_html(keep_len=keep_len)
         # print("The html content: ", html_content)
 
diff --git a/metagpt/tools/libs/editor.py b/metagpt/tools/libs/editor.py
index e032dcef5..78560e375 100644
--- a/metagpt/tools/libs/editor.py
+++ b/metagpt/tools/libs/editor.py
@@ -100,7 +100,7 @@ class Editor:
                     file_path=file_path,
                     block_content=block_content,
                 )
-                self.resource.report(result.file_path, "path")
+                self.resource.report(result.file_path, "path", extra={"type": "search", "line": i, "symbol": symbol})
                 return result
         return None
 
diff --git a/metagpt/tools/libs/git.py b/metagpt/tools/libs/git.py
index eb3fd6822..b4d759bf4 100644
--- a/metagpt/tools/libs/git.py
+++ b/metagpt/tools/libs/git.py
@@ -9,7 +9,6 @@ from github.Issue import Issue
 from github.PullRequest import PullRequest
 
 from metagpt.tools.tool_registry import register_tool
-from metagpt.utils.git_repository import GitBranch, GitRepository
 
 
 @register_tool(tags=["software development", "git", "Commit the changes and push to remote git repository."])
@@ -18,7 +17,7 @@ async def git_push(
     access_token: str,
     comments: str = "Commit",
     new_branch: str = "",
-) -> GitBranch:
+) -> "GitBranch":
     """
     Pushes changes from a local Git repository to its remote counterpart.
 
@@ -49,6 +48,8 @@ async def git_push(
         base branch:'master', head branch:'feature/new', repo_name:'iorisa/snake-game'
 
     """
+    from metagpt.utils.git_repository import GitRepository
+
     if not GitRepository.is_git_dir(local_path):
         raise ValueError("Invalid local git repository")
 
diff --git a/metagpt/utils/common.py b/metagpt/utils/common.py
index 9e9bb034c..7303d1f47 100644
--- a/metagpt/utils/common.py
+++ b/metagpt/utils/common.py
@@ -646,7 +646,7 @@ def role_raise_decorator(func):
             raise Exception(format_trackback_info(limit=None))
         except Exception as e:
             if self.latest_observed_msg:
-                logger.warning(
+                logger.exception(
                     "There is a exception in role's execution, in order to resume, "
                     "we delete the newest role communication message in the role's memory."
                 )
diff --git a/metagpt/utils/file.py b/metagpt/utils/file.py
index f62b44eb8..a8ed482d9 100644
--- a/metagpt/utils/file.py
+++ b/metagpt/utils/file.py
@@ -9,6 +9,7 @@
 from pathlib import Path
 
 import aiofiles
+from fsspec.implementations.memory import MemoryFileSystem as _MemoryFileSystem
 
 from metagpt.logs import logger
 from metagpt.utils.exceptions import handle_exception
@@ -68,3 +69,10 @@ class File:
             content = b"".join(chunks)
             logger.debug(f"Successfully read file, the path of file: {file_path}")
             return content
+
+
+class MemoryFileSystem(_MemoryFileSystem):
+
+    @classmethod
+    def _strip_protocol(cls, path):
+        return super()._strip_protocol(str(path))
diff --git a/metagpt/utils/parse_html.py b/metagpt/utils/parse_html.py
index 65aa3f236..3aac8ca6c 100644
--- a/metagpt/utils/parse_html.py
+++ b/metagpt/utils/parse_html.py
@@ -7,6 +7,8 @@ from urllib.parse import urljoin, urlparse
 from bs4 import BeautifulSoup
 from pydantic import BaseModel, PrivateAttr
 
+import htmlmin
+
 
 class WebPage(BaseModel):
     inner_text: str
@@ -38,6 +40,22 @@ class WebPage(BaseModel):
             elif url.startswith(("http://", "https://")):
                 yield urljoin(self.url, url)
 
+    def get_slim_soup(self, keep_links: bool = False):
+        soup = _get_soup(self.html)
+        keep_attrs = ["class"]
+        if keep_links:
+            keep_attrs.append("href")
+
+        for i in soup.find_all(True):
+            for name in list(i.attrs):
+                if i[name] and name not in keep_attrs:
+                    del i[name]
+
+        for i in soup.find_all(["svg", "img", "video", "audio"]):
+            i.decompose()
+
+        return soup
+
 
 def get_html_content(page: str, base: str):
     soup = _get_soup(page)
@@ -48,7 +66,12 @@ def get_html_content(page: str, base: str):
 def _get_soup(page: str):
     soup = BeautifulSoup(page, "html.parser")
     # https://stackoverflow.com/questions/1936466/how-to-scrape-only-visible-webpage-text-with-beautifulsoup
-    for s in soup(["style", "script", "[document]", "head", "title"]):
+    for s in soup(["style", "script", "[document]", "head", "title", "footer"]):
         s.extract()
 
     return soup
+
+
+def simplify_html(html: str, url: str, keep_links: bool = False):
+    html = WebPage(inner_text="", html=html, url=url).get_slim_soup(keep_links).decode()
+    return htmlmin.minify(html, remove_comments=True, remove_empty_space=True)
diff --git a/metagpt/utils/report.py b/metagpt/utils/report.py
index a61c77381..2d72af111 100644
--- a/metagpt/utils/report.py
+++ b/metagpt/utils/report.py
@@ -39,6 +39,7 @@ class BlockType(str, Enum):
     GALLERY = "Gallery"
     NOTEBOOK = "Notebook"
     DOCS = "Docs"
+    THOUGHT = "Thought"
 
 
 END_MARKER_NAME = "end_marker"
@@ -55,23 +56,23 @@ class ResourceReporter(BaseModel):
     callback_url: str = Field(METAGPT_REPORTER_DEFAULT_URL, description="The URL to which the report should be sent")
     _llm_task: Optional[asyncio.Task] = PrivateAttr(None)
 
-    def report(self, value: Any, name: str):
+    def report(self, value: Any, name: str, extra: Optional[dict] = None):
         """Synchronously report resource observation data.
 
         Args:
             value: The data to report.
             name: The type name of the data.
         """
-        return self._report(value, name)
+        return self._report(value, name, extra)
 
-    async def async_report(self, value: Any, name: str):
+    async def async_report(self, value: Any, name: str, extra: Optional[dict] = None):
         """Asynchronously report resource observation data.
 
         Args:
             value: The data to report.
             name: The type name of the data.
         """
-        return await self._async_report(value, name)
+        return await self._async_report(value, name, extra)
 
     @classmethod
     def set_report_fn(cls, fn: Callable):
@@ -100,20 +101,20 @@ class ResourceReporter(BaseModel):
         """
         cls._async_report = fn
 
-    def _report(self, value: Any, name: str):
+    def _report(self, value: Any, name: str, extra: Optional[dict] = None):
         if not self.callback_url:
             return
 
-        data = self._format_data(value, name)
+        data = self._format_data(value, name, extra)
         resp = requests.post(self.callback_url, json=data)
         resp.raise_for_status()
         return resp.text
 
-    async def _async_report(self, value: Any, name: str):
+    async def _async_report(self, value: Any, name: str, extra: Optional[dict] = None):
         if not self.callback_url:
             return
 
-        data = self._format_data(value, name)
+        data = self._format_data(value, name, extra)
         url = self.callback_url
         _result = urlparse(url)
         sessiion_kwargs = {}
@@ -129,9 +130,16 @@ class ResourceReporter(BaseModel):
                 resp.raise_for_status()
                 return await resp.text()
 
-    def _format_data(self, value, name):
+    def _format_data(self, value, name, extra):
         data = self.model_dump(mode="json", exclude=("callback_url", "llm_stream"))
-        data["value"] = str(value) if isinstance(value, Path) else value
+        if isinstance(value, BaseModel):
+            value = value.model_dump(mode="json")
+        elif isinstance(value, Path):
+            value = str(value)
+
+        if name == "path":
+            value = os.path.abspath(value)
+        data["value"] = value
         data["name"] = name
         role = CURRENT_ROLE.get(None)
         if role:
@@ -139,6 +147,8 @@ class ResourceReporter(BaseModel):
         else:
             role_name = os.environ.get("METAGPT_ROLE")
         data["role"] = role_name
+        if extra:
+            data["extra"] = extra
         return data
 
     def __enter__(self):
@@ -252,6 +262,16 @@ class TaskReporter(ObjectReporter):
     block: Literal[BlockType.TASK] = BlockType.TASK
 
 
+class ThoughtReporter(ObjectReporter):
+    """Reporter for object resources to Task Block."""
+
+    block: Literal[BlockType.THOUGHT] = BlockType.THOUGHT
+
+    async def __aenter__(self):
+        await self.async_report({})
+        return await super().__aenter__()
+
+
 class FileReporter(ResourceReporter):
     """File resource callback for reporting complete file paths.
 
@@ -259,13 +279,23 @@ class FileReporter(ResourceReporter):
     if the file can be partially output for display first, use streaming callback.
     """
 
-    def report(self, value: Union[Path, dict, Any], name: Literal["path", "meta", "content"] = "path"):
+    def report(
+        self,
+        value: Union[Path, dict, Any],
+        name: Literal["path", "meta", "content"] = "path",
+        extra: Optional[dict] = None,
+    ):
         """Report file resource synchronously."""
-        return super().report(value, name)
+        return super().report(value, name, extra)
 
-    async def async_report(self, value: Path, name: Literal["path", "meta", "content"] = "path"):
+    async def async_report(
+        self,
+        value: Union[Path, dict, Any],
+        name: Literal["path", "meta", "content"] = "path",
+        extra: Optional[dict] = None,
+    ):
         """Report file resource asynchronously."""
-        return await super().async_report(value, name)
+        return await super().async_report(value, name, extra)
 
 
 class NotebookReporter(FileReporter):
diff --git a/requirements.txt b/requirements.txt
index b40c69c9f..83a904156 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -71,4 +71,6 @@ dashscope==1.14.1
 rank-bm25==0.2.2  # for tool recommendation
 gymnasium==0.29.1
 pylint~=3.0.3
-pygithub~=2.3
\ No newline at end of file
+pygithub~=2.3
+htmlmin
+fsspec