Pr Ship

Pre-ship risk report for OpenClaw PRs. Dynamically explores the codebase to assess module risk, blast radius, and version-specific gotchas. Scores each findi...

What to check before installing or running this skill: - Review scripts/test-update-pipeline.sh line-by-line. It contains hardcoded paths (/home/dev/.openclaw/skills/pr-ship, $HOME/.openclaw/cron/jobs.json) and will read local cron config and git metadata — ensure those checks are acceptable for your environment. - The SKILL.md and playbook assume many command-line tools (git, grep, sed, node, jq, python3, pnpm, clawhub). The skill does not declare these; make sure your environment provides them and consider running in a sandbox or a dedicated dev container. - The skill's update flow (clawhub update, cron-driven CURRENT-CONTEXT.md refresh) will perform network git operations (git fetch upstream). Confirm the remote origins are trustworthy and that no credentials will be accidentally exposed or used for unintended pushes. - If you plan to enable automated updates (cron/clawhub), run the update script manually first in a safe environment and verify it does not perform unexpected writes or network pushes (the test script explicitly warns cron should not run git push or clawhub publish automatically). - The skill is coherent with its stated purpose, but because it touches system files and assumes an update pipeline, treat it as potentially impactful: run reviews locally, verify remote provenance (the package.json points to a GitHub repo), and avoid enabling automatic cron updates until you confirm the workflow and remotes. If you want, I can extract the specific lines from the script that read host paths and network operations, or produce a short checklist of commands to run locally to validate behavior safely.

Type: OpenClaw Skill Name: pr-ship Version: 1.0.13 The skill is designed for local codebase analysis using explicitly read-only commands (`grep`, `find`, `ls`, `git diff`). The `SKILL.md` and `EXPLORATION-PLAYBOOK.md` contain strong guardrails, explicitly instructing the AI agent *not* to execute commands that modify files or perform build/test actions, but only to recommend them to the user. The `scripts/test-update-pipeline.sh` script, while using more powerful commands, is for testing the skill's update mechanism and includes checks that *prevent* automated `git push` or `clawhub publish` in the cron job. A 'Security Notice' in `SKILL.md` warns users about potential secret exposure in generated reports, which is a transparency measure, not an instruction for malicious exfiltration. No evidence of intentional harmful behavior (e.g., data exfiltration, persistence, unauthorized remote control) was found.