Secret Scan

SkillDev tools

Scan the vault and repo for leaked secrets — API keys, tokens, private keys — before committing or sending content externally. Use before a commit/push, before posting vault content to any external surface, or when the user asks to check for exposed credentials.

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the Secret Scan skill

What this skill tells your AI

The instructions your AI receives, as published by thesmokedev/taskchad-os in optional-skills/security/secret-scan/SKILL.md and read by ahel’s review.

Catch credentials before they leak into a commit or an outbound message.

Run

uv run python optional-skills/security/secret-scan/scripts/scan.py .
# scan a specific path before sending it out:
uv run python optional-skills/security/secret-scan/scripts/scan.py vault/notes/draft.md

Exit code is non-zero if anything matches, so it drops into a pre-commit hook or a pre-send guard.

What it flags

Common high-signal patterns: AWS keys, Google API keys, Slack/Telegram/GitHub tokens, OpenAI/Anthropic keys, private-key PEM blocks, and generic SECRET/PASSWORD/TOKEN = "..." assignments. The detector lives in scripts/scan.py — extend the pattern table there as new providers appear.

Where this belongs in the framework

This is the read-only half of the framework's default-deny mutation policy. Before any skill sends vault content to an external surface, run this scan on the exact payload. A match means stop and ask the user, never auto-send. The scan never edits or redacts on its own — it reports; a human decides.

Reducing false positives

  • Respect .gitignore / .graphifyignore; skip .env*, node_modules, .git.
  • Example/placeholder values (xxxx, your-key-here, example) are downranked.
  • When in doubt, surface the finding with file:line and let the user judge.

Signals

GitHub stars
23
Forks
5
Last commit
Sep 2026

ahel review

  • K6low
    bundled executables the agent is told to run

Automated review, not a security audit. Ruleset v1+k2.

Advanced
Catalog kind
skill
Gateway key
secret-scan-thesmokedev
Source
github.com/thesmokedev/taskchad-os