Change log
New and updated fixes
31 guides added or changed status in the last 7 days. A first status check is listed but not counted as an update.
· 31
- Guide added Coding agent CLI Linux sandbox fails with "mountinfo path is not absolute" after an update (Coding agent CLI)
- Guide added Coding agent CLI VS Code extension drops submitted messages or leaves them queued (Coding agent CLI)
- Guide added Coding agent CLI on Windows: sandbox setup fails with a sharing violation, every command fails (Coding agent CLI)
- Guide added Coding agent CLI Desktop "Run agent in WSL": every command fails with "Failed to create unified exec process" (Coding agent CLI)
- Guide added Coding agent CLI: "API Error: Connection closed mid-response" (Coding agent CLI)
- Guide added Coding agent ignores "stop writing comments" rules in its project instructions file (Coding agent CLI)
- First status check Coding agent stuck on 'did not conform to the edit format' Workaround source
- First status check Adding a second agent to the quick-start makes it loop forever Workaround source
- First status check Graph flow errors with 'unable to find next speaker' when a loop has tools Workaround source
- First status check 'str' object has no attribute 'get' when the chat finishes Fixed (verified) source
- First status check Browser agent can't pick an option from a dropdown Fixed (verified) source
- First status check Browser agent keeps going after it already called done Workaround source
- First status check Browser agent keeps emitting empty actions until max steps Workaround source
- First status check Click on a missing element is reported as success Open source
- First status check Every tool call arrives with empty arguments and fails validation Workaround source
- First status check Error messages name the symptom, not the cause Workaround source
- First status check JSON output parser returns None instead of raising on bad JSON Workaround source
- First status check Parser fails when the model wraps JSON in ```JSON instead of ```json Workaround source
- First status check Agent repeats the same tool call until it hits the recursion limit Workaround source
- First status check Cancelled async tool leaves history invalid even with handle_tool_errors=True Open source
- First status check Token limit exceeded because retriever output piles up in chat history Workaround source
- First status check Setting recursion_limit to 10000 silently does nothing Workaround source
- First status check "tool_use ids were found without tool_result blocks" on a saved thread Workaround source
- First status check Streamable HTTP client keeps reconnecting past its retry limit Open source
- First status check Fetch tool can be steered to localhost or cloud metadata Fixed (verified) source
- First status check Filesystem server says 'path outside allowed directories' on the first call Fixed (verified) source
- First status check npx-based server shows zero tools when Node comes from nvm Workaround source
- First status check Stateless MCP server logs ERROR every time a client disconnects Workaround source
- First status check One bad tool call ends the whole agent run Open source
- First status check AI Agent tools fail with 'schema must be type object, got None' after upgrading Workaround source
- First status check Tool calls fail because arguments come back as XML-like text Workaround source
· 25
- Guide added Coding agent stuck on 'did not conform to the edit format' (Aider)
- Guide added Adding a second agent to the quick-start makes it loop forever (AutoGen)
- Guide added Graph flow errors with 'unable to find next speaker' when a loop has tools (AutoGen)
- Guide added 'str' object has no attribute 'get' when the chat finishes (AutoGen)
- Guide added Browser agent can't pick an option from a dropdown (browser-use)
- Guide added Browser agent keeps going after it already called done (browser-use)
- Guide added Browser agent keeps emitting empty actions until max steps (browser-use)
- Guide added Click on a missing element is reported as success (browser-use)
- Guide added Every tool call arrives with empty arguments and fails validation (CrewAI)
- Guide added Error messages name the symptom, not the cause (Hermes Agent)
- Guide added JSON output parser returns None instead of raising on bad JSON (LangChain)
- Guide added Parser fails when the model wraps JSON in ```JSON instead of ```json (LangChain)
- Guide added Agent repeats the same tool call until it hits the recursion limit (LangGraph)
- Guide added Cancelled async tool leaves history invalid even with handle_tool_errors=True (LangGraph)
- Guide added Token limit exceeded because retriever output piles up in chat history (LangGraph)
- Guide added Setting recursion_limit to 10000 silently does nothing (LangGraph)
- Guide added "tool_use ids were found without tool_result blocks" on a saved thread (LangGraph)
- Guide added Streamable HTTP client keeps reconnecting past its retry limit (MCP Python SDK)
- Guide added Fetch tool can be steered to localhost or cloud metadata (MCP)
- Guide added Filesystem server says 'path outside allowed directories' on the first call (MCP)
- Guide added npx-based server shows zero tools when Node comes from nvm (MCP)
- Guide added Stateless MCP server logs ERROR every time a client disconnects (MCP Python SDK)
- Guide added One bad tool call ends the whole agent run (n8n)
- Guide added AI Agent tools fail with 'schema must be type object, got None' after upgrading (n8n)
- Guide added Tool calls fail because arguments come back as XML-like text (Continue)