Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
What This Skill Does
Runs an autonomous code review loop against a GitHub pull request using Greptile. The agent fetches review feedback, applies fixes, pushes changes, and re-triggers the review until the score reaches 4/5 or the round limit is hit. Architectural concerns and stalled reviews are escalated to a human.
Eliminates the manual loop of reading feedback, switching to an editor, committing fixes, and re-requesting review by handling all of it autonomously.
When to use it
- Auto-fixing Greptile review comments on a freshly opened PR
- Merging a PR automatically once it passes the quality threshold
- Iterating through multiple review rounds without manual intervention
- Escalating design pattern or architecture concerns to a human
- Tracking review progress across concurrent PRs in a repository
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: pr review owner/repo#42
- 1Fetch the latest Greptile review and inline comments via GitHub API
- 2Parse the score from the review body
- 3Apply fixes to each referenced file and commit with a descriptive message listing each change
- 4Push the branch and comment '@greptileai review' to re-trigger the review
- 5Repeat until score is 4/5 or max rounds reached, then merge or escalate to human
PR merged with all Greptile feedback addressed, or escalation message sent if architectural changes are required
Requirements
Accounts, API keys, or tools you or your AI assistant may need to set up while using this skill.