Peter Pr Ops

PR 收口自动化。单 PR 串行合并与多 PR 批量清理，减少重复人工操作。

Before installing or enabling this skill: (1) Confirm the host has git and the GitHub CLI (gh) available; the skill expects them even though the manifest does not declare them. (2) Inspect the repository scripts referenced (scripts/automerge, scripts/massageprs, scripts/ensure-workflow-docs) and any script at $HOME/ai_code/study_peter/scripts — they will be executed and can run arbitrary commands. (3) Understand which GitHub credentials/gh auth are used and restrict their scopes (use a least-privilege token or run in a test/fork environment). (4) Test the skill on a non-critical repository or a fork to confirm behavior. (5) Prefer the author update the manifest to list required binaries (git, gh) and document required auth scopes and exact file paths to avoid hidden execution of unexpected scripts. If you cannot review the scripts or do not trust them, do not run this skill with an account that has merge permissions.

Type: OpenClaw Skill Name: peter-pr-ops Version: 1.2.1 The `SKILL.md` file instructs the AI agent to execute local scripts, most notably `"$repo_root/scripts/ensure-workflow-docs"` (and also `scripts/automerge`, `scripts/massageprs`). The `repo_root` is determined by `git rev-parse --show-toplevel`. This creates a severe remote code execution (RCE) vulnerability, as an attacker could craft a malicious git repository containing a harmful `scripts/ensure-workflow-docs` script, which the agent would then execute when operating within that repository. This is classified as suspicious due to the high-risk RCE vulnerability, even though the skill bundle itself does not contain explicit malicious payloads.