deepagents
https://github.com/langchain-ai/deepagents
Python
Deep Agents is an agent harness built on langchain and langgraph. Deep Agents are equipped with a planning tool, a filesystem backend, and the ability to spawn subagents - making them well-equipped to handle complex agentic tasks.
Triage Issues!
When you volunteer to triage issues, you'll receive an email each day with a link to an open issue that needs help in this project. You'll also receive instructions on how to triage issues.
Triage Docs!
Receive a documented method or class from your favorite GitHub repos in your inbox every day. If you're really pro, receive undocumented methods or classes and supercharge your commit history.
Python not yet supported1 Subscribers
Add a CodeTriage badge to deepagents
Help out
- Issues
- Summarization overflow fallback (`_slice_read_file_tm`) silently drops image blocks from `read_file` tool results
- Native Communication and Coordination Between Peer-Level Deep Agents
- Approval policy file for fine-grained auto-approval rules
- Guest `try/catch` around `task()` swallows a propagated `GraphInterrupt`, so a HITL gate inside a dispatched subagent is silently skipped
- Resuming a HITL interrupt raised inside `eval` re-executes the whole eval body against the already-mutated REPL
- RFC: converge `read_file` offset on 1-indexed line numbers (input + output)
- Give offloaded tool-result files a recognizable text extension
- `ModelCallLimitMiddleware` + `response_format` ends with empty `structured_response`
- Add `ToolSelectionMiddleware` for per-turn tool filtering
- MCP-prefixed tool names can collide with MCP and built-in tools
- Docs
- Python not yet supported