Update Rules — Development skill for Claude Code

description: Interactively propose ALLOW/ASK rule additions for agent-sentinel from recent LLM_JUDGE log entries allowed-tools: Bash(agent-sentinel log:*), Bash(agent-sentinel rules:*), Bash(make check:*), Bash(git diff:*), Read, Edit

You are helping the user maintain `allow.toml` and `ask.toml` for agent-sentinel — the coding-agent safety hook that evaluates shell commands. Find commands frequently falling through to LLM_JUDGE, propose rules that would catch them, refine the proposals