loop-on-ci

An agent skill by cursor, from cursor/plugins. Tags: automation, ci-cd, debugging, github-actions, monitoring.

What it does

Monitor PR checks and fix failures until green. Uses gh pr checks as the source of truth for PR-attached checks.

Install

With the skills CLI, which installs into Claude Code, Codex, Cursor and other agents:

npx skills add cursor/plugins --skill loop-on-ci

Or copy the skill folder into Claude Code's skills directory by hand (~/.claude/skills for every project, or .claude/skills inside one):

git clone --depth 1 https://github.com/cursor/plugins
cp -r plugins/cursor-team-kit/skills/loop-on-ci ~/.claude/skills/loop-on-ci

Safety box score

Grade B, low risk: 25 of 100 risk points, rated 2026-10-16. The grade rates what the skill and its scripts can reach on the machine of whoever installs it, across eight categories from shell execution to secrets access. How the score works.

This skill is a GitHub-CLI loop for monitoring PR checks and reading failing CI logs until the PR is green. Its main risks are authenticated network access to GitHub, exposure of private PR/check data through logs, and an open-ended retry loop that can keep the agent working until success.

Category levels, from 0 (none) to 3 (high)
Category	Level	Why
Code & shell execution	1 (low)	It only runs plain `gh` CLI commands like `gh pr view` and `gh pr checks`; there is no code evaluation, downloaded installer, or `eval`/`curl \| sh` behavior.
Network access	2 (moderate)	The skill repeatedly contacts GitHub for PR metadata, check status, and logs via `gh pr view`, `gh pr checks --watch --fail-fast`, and `gh run view <run-id> --log-failed`.
Filesystem reach	0 (none)	The file contains no explicit read/write/delete operations on local paths, dotfiles, or system locations.
Secrets & credentials	2 (moderate)	It runs under the user's authenticated `gh` session and can access private PR data and failed CI logs, which may contain sensitive output from the repo or build environment.
Privilege & persistence	0 (none)	There is no sudo, service installation, shell startup editing, or other privilege-changing step.
Instruction hijack surface	0 (none)	The only potentially concerning text is a prohibition: `Do not bypass hooks (--no-verify) to force progress.` It does not instruct the agent to ignore rules or follow external instructions.
Irreversible actions	1 (low)	The workflow is a monitoring/retry loop that says to `repeat until green`, but it does not itself merge, deploy, publish, or send messages.
Transparency	0 (none)	All content is readable text; there are no binaries, base64 blobs, or unreviewable installers.

Before installing

Run `gh auth status` and confirm the account/repo scopes are what you expect.
Try it first on a non-sensitive PR and inspect `gh run view --log-failed` output for secret leakage.
Decide whether you are comfortable with an agent repeatedly polling GitHub until checks pass.

Rated 2026-10-16 by an automated review of 1 file from the skill's repository.

Source

Repository: cursor/plugins (all skills from this repository)
Path: cursor-team-kit/skills/loop-on-ci/SKILL.md
Branch: main
Collection: cursor
Updated: 2026-09-23

fix-ci — Find failing PR checks, inspect logs or external check links, and apply focused fixes
control-cli — Build or adapt a local harness to drive, inspect, and profile an interactive CLI or TUI without external services.
control-ui — Build or adapt a local browser/CDP harness to drive and inspect a web, IDE, or Electron UI.
reproduce-and-fix-issues — Reproduce triaged Slack bugs through a configured app-control adapter, verify existing fixes.
agentic-actions-auditor — Audits GitHub Actions workflows for security vulnerabilities in AI agent integrations including Claude Code Action, Gemini CLI, OpenAI Codex.
agents-deploy — Use when deploying your agent to AWS, or when a deploy has failed.