megalinter
Entry point for everything MegaLinter. Use when the user wants to lint their repository, set up MegaLinter, check or fix lint errors, make CI lint jobs pass, or says "run megalinter", "fix lint errors", "make the linters happy", "clean up code quality". Detects the repository state and orchestrates the megalinter-setup, megalinter-check and megalinter-fix skills in a loop until the repository is clean.
How do I install this agent skill?
npx skills add https://github.com/oxsecurity/megalinter --skill megalinterIs this agent skill safe to install?
- Gen Agent Trust Hubpass
The MegaLinter orchestrator skill automates code quality checks by coordinating linter execution, applying fixes, and monitoring CI jobs. It uses standard tools like npx and Git to manage the linting workflow. The behavior described is consistent with the legitimate operation of the MegaLinter toolset by OX Security, and no malicious patterns were identified.
- Socketpass
No alerts
- Snykwarn
Risk: MEDIUM · 1 issue
What does this agent skill do?
MegaLinter orchestrator
You orchestrate MegaLinter on this repository. MegaLinter is a mega-linter aggregating 100+ linters, running as a Docker image (locally via npx mega-linter-runner) or as a CI job.
Workflow
- Detect repository state:
- No
.mega-linter.ymland no MegaLinter CI workflow file → run themegalinter-setupskill first. - Configuration present → go to step 2.
- No
- Collect errors with the
megalinter-checkskill:- If a MegaLinter CI job is currently running or just failed for the current branch/PR → use its watch mode.
- Otherwise → use its local mode (requires a container engine — docker or podman; the skill handles asking the
user to install/start one if missing, and offers native mode (beta, linters installed on the machine, no
container) when no engine is wanted; native mode only runs the linters listed in
NATIVE_LINTERS). Local runs are resource-consuming (MegaLinter is Docker-based, needs a good computer configuration and a good internet connection, and can download images of several GB), so when a CI run is also achievable (the repository has a MegaLinter CI workflow — e.g. by pushing the branch or opening a PR), using CI is usually recommended: ask the user which one they want (structured question when your platform has one, CI first/recommended) instead of defaulting to local.
- If errors were found, run the
megalinter-fixskill with the collected error list. - Re-check: after fixes, run
megalinter-checkagain — in local mode, prefer its targeted re-check (parallel standalone linter runs restricted to previously-failing linters and fixed files). - Repeat steps 3-4 at most 3 times. If errors remain after 3 iterations, stop and report the remaining errors with your recommendation (fix manually, disable rules, or disable linters).
- Performance: if
megalinter-checkreportedslow_linters(even on a green run), relay its performance suggestions to the user at the end of the loop — speed wins are proposed, never applied without agreement.
Rules
- Never commit or push on the default branch (
main/master): create a branch likemegalinter/fix-<topic>first. - Ask the user before disabling any linter or rule, and before pushing commits.
- Never force-push, with one exception: when MegaLinter itself pushed a
[MegaLinter] Apply linters fixescommit onto the watched branch,megalinter-checkamends it with a 🤖 prefix and re-pushes it with--force-with-leaseto re-trigger the CI checks (see its "MegaLinter auto-fix commits" section for the conditions). Never use plain--force, and never force-push in any other situation. - Keep the user informed with a short status after each phase (errors found, fixes applied, remaining).
Optimization: sub-agents (Claude Code and compatible agents)
If your environment supports spawning sub-agents (e.g. a Task/Agent tool) and the megalinter-watcher, megalinter-runner, megalinter-fixer agent definitions are available:
- Installed as an agent plugin (Claude Code, Cursor, Copilot), they ship with the plugin. Some platforms list them under a namespaced name (
megalinter:megalinter-watcher), others under the bare name. - Installed as skills, the
megalinter-setupskill copies them into your platform's agents folder (e.g..claude/agents/,.opencode/agent/,.github/agents/where the file name ends with.agent.md) under their bare names.
Use whichever of the two forms your platform actually lists, and fall back to inline execution if neither is available.
Then:
- Delegate CI watching to
megalinter-watcherand local runs tomegalinter-runner— they keep the large logs out of your context and return only a compact error list. - Fan out one
megalinter-fixerper failing linter to fix several linters in parallel.
If sub-agents are not available, perform every step inline following the skill instructions.
How can the creator link this skill?
Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.
<a href="https://skillzs.dev/skills/oxsecurity/megalinter/megalinter">View megalinter on skillZs</a>