agentsclimarketplace

Api key leak checker leop

Skill leo-cheung-itlger/api-key-leak-checker-leop/skills/api-key-leak-checker-leop

A Codex skill and pre-publish gate for checking API key leaks before open-sourcing projects.

Install
npx -y skills add leo-cheung-itlger/api-key-leak-checker-leop --skill api-key-leak-checker-leop

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 3 stars3 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Check GitHub repositories, local Git projects, AI coding projects, and Vibe Coding apps for possible API key, token, secret, .env, and config leaks. Use when the user asks to scan for leaked keys, exposed credentials, GitHub secret scanning, Gitleaks, TruffleHog, .gitignore hardening, push protection, API key rotation, or remediation after accidentally committing secrets.

SKILL.md

4.7 KB, as published. Nobody here has run it

API Key Leak Checker LEOP

Purpose

Use this skill to run a safety-first API key leak check, remediation workflow, and pre-publish gate for open-source releases. Do not build a custom scanner from scratch unless existing tools are unavailable. Prefer existing tools, project scripts, GitHub built-ins, and provider dashboards.

Treat any secret that reached a public repository, public issue, public build log, shared screenshot, or deployed client bundle as compromised. Removing the file is not enough; the key must be revoked or rotated at the provider.

Fast Workflow

  1. Identify the target: local repo path, GitHub URL, owner/repo, username, or organization.
  2. Preserve safety: never ask the user to paste secrets, passwords, cookies, tokens, or raw credentials into chat.
  3. Inspect repo state before scanning:
    • git status --short
    • git remote -v
    • .gitignore, .env*, config*.json, *.key, *.pem, CI files, deployment files
  4. Run existing scanners first:
    • Run scripts/pre_publish_check.ps1 -Path <repo> as the local gate
    • Prefer gitleaks detect --source . --redact --no-banner
    • Use trufflehog git file://. --only-verified --no-update when available
    • For GitHub exposure, use gh and GitHub web/secret scanning when authenticated
  5. Classify findings:
    • Critical: likely live secret in public repo, Git history, public build logs, or client-side bundle
    • High: likely secret in private repo history or shared artifact
    • Medium: secret-like value, placeholder risk, weak .gitignore, or missing push protection
    • Low: docs/example-only placeholder with clear fake marker
  6. Remediate in the right order:
    • Tell the user to revoke or rotate exposed keys in the provider dashboard first.
    • Remove hardcoded secrets from code and config.
    • Move runtime values to environment variables or a secret manager.
    • Add ignore rules and pre-commit/CI scanning.
    • Only discuss Git history rewriting after rotation, and require explicit user approval before destructive or force-push steps.
  7. Report clearly:
    • What was checked
    • What was found
    • Whether any key must be rotated
    • What changed locally
    • What the user must still do manually

Pre-Publish Gate

Before creating a public repository, pushing code, opening a source release, or sharing a project zip, run:

powershell -ExecutionPolicy Bypass -File "C:\Users\13678\.codex\skills\api-key-leak-checker-leop\scripts\pre_publish_check.ps1" -Path "<repo-path>"

Use -Strict when the user wants medium-risk filenames to block publication too:

powershell -ExecutionPolicy Bypass -File "C:\Users\13678\.codex\skills\api-key-leak-checker-leop\scripts\pre_publish_check.ps1" -Path "<repo-path>" -Strict

Interpret exit codes:

  • 0: no blocking indicators found
  • 2: block publish until findings are reviewed and fixed
  • other nonzero: check failed and must be rerun after fixing the error

The script redacts values and reports rule names, files, and line numbers only. Do not publish when it reports Critical or High.

GitHub CLI Login

If gh auth status says not logged in, use browser login and verify persistence:

gh auth login --hostname github.com --git-protocol https --web
gh auth setup-git
gh auth status
Test-Path "$env:APPDATA\GitHub CLI\hosts.yml"

If browser login does not persist, guide the user to create a GitHub Personal Access Token in the browser and enter it locally with gh auth login --with-token. Do not ask them to paste the token into chat or write it into files.

If the user already has GITHUB_TOKEN but gh asks for login, map it to GH_TOKEN in the user environment and verify with gh api user --jq '.login'. Do not print the token value.

Tool References

Read references/tooling.md when choosing commands, installing missing scanners, or diagnosing authentication.

Read references/remediation.md when a finding looks real, when a key may have entered history, or when the user asks what to do after leakage.

Output Template

## API Key Leak Check

Checked:
- Local repo:
- Remote:
- Tools:

Findings:
- Critical:
- High:
- Medium:
- Low:

Required user actions:
- Rotate/revoke:
- Provider dashboards:

Local hardening:
- .gitignore:
- Environment variables:
- Pre-commit/CI:

Residual risk:
- Git history:
- Public forks/caches:

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.