Feedback wanted: what guardrail is missing before running agents unattended?

[rxdt]

Ralph intentionally keeps the loop small: specs, fresh context, status handoff docs, git hooks, CI gates, and hard iteration/time caps.

I want feedback from people actually running Claude Code, Codex CLI, or Gemini CLI on real repos.

Questions:

• What guardrail would you require before letting an agent run for multiple iterations?
• Is this too strict, not strict enough, or just missing the obvious thing?
• Where should Ralph stay minimal instead of growing into an orchestrator?

Try it from GitHub:

uvx --from git+https://github.com/rxdt/py_ralph_frame ralph-harness demo

Longer writeup:
https://gist.github.com/rxdt/6caddb5c829c3ef5bc2c7881c441e61b