Optional 30–60 minute research pilot on coding-agent review evidence

[kodlbegiko]

Hello tmuxp maintainers,

I maintain Agent Completion Ledger (ACL), a small open-source research tool that checks a coding-agent completion claim against repository evidence selected by a maintainer. I am testing an unproven research question: whether a structured ACL report improves a non-author maintainer's decision compared with the same agent summary alone.

tmuxp appears suitable because its contributor workflow already makes the review evidence explicit through ruff, mypy, pytest, and repository agent guidance. A pilot could use one bounded parser or configuration-validation case with public fixtures, or you can review the fixed ten-task pack without installing anything.

Expected time is 30–60 minutes. Static-only --no-exec review is available, and repository integration is optional. I will not request shell history, personal tmux configuration, private code, credentials, or secrets. I will not automatically create a PR in tmuxp.

The result may support, narrow, or reject ACL's value claim; participation is not endorsement, and ACL is not an industry standard. You may decline or stop at any time, and a decline ends contact.

Would you be willing to inspect the fixed materials before deciding?

Repository: https://github.com/kodlbegiko/agent-completion-ledger

Protocol: https://github.com/kodlbegiko/agent-completion-ledger/blob/main/docs/EXTERNAL-VALIDATION-PROTOCOL.md