mini-code-agent-langgraph 0.3.2: launch notes and technical feedback

[wusuiling-if]

mini-code-agent-langgraph 0.3.2 is now available on PyPI. It is a compact, inspectable LangGraph coding-agent CLI with verification-bound submission, redacted resumable trajectories, sandboxed tools, and HMAC-authenticated conflict-aware Undo.

Try the deterministic local flow without an API key:

python -m pip install mini-code-agent-langgraph==0.3.2
mca demo

Run mca demo on macOS, Linux, or WSL2. Native Windows supports informational CLI and configuration paths only.

The v0.3.2 offline suite passes 11/11 deterministic runtime-policy cases: nine verified submissions and two expected refusals, with zero unexpected submissions and zero unrelated changes. This is policy-conformance evidence using scripted local decisions—not model-quality, autonomous-repair, provider, real-world success, or SWE-bench evidence.

Useful technical feedback includes:

• Is the verification boundary understandable and strict enough?
• Is the trajectory/resume model inspectable enough to debug failures?
• Are the Undo conflict and authentication semantics clear?
• Are sandbox setup and /ask versus /code ergonomics practical?
• Which LangChain/LangGraph coupling would you remove first?

Please use Issues for reproducible bugs or scoped feature requests, and avoid posting credentials, private source, raw trajectories, or Undo journals.

• Release: https://github.com/wusuiling-if/mini-code-agent-langgraph/releases/tag/v0.3.2
• PyPI: https://pypi.org/project/mini-code-agent-langgraph/0.3.2/