Agent Sentinel

Scan OpenClaw/Hermes skills for risky permission patterns before installation

Install

openclaw skills install @itspremkumar/agent-sentinel

Skill Security Sentinel

Scan OpenClaw/Hermes skills for risky permission patterns before you install them.

Keywords: security, audit, skill, openclaw, hermes, vetting, cli, python, open-source, agent, automation, MIT

Part of the itsPremkumar Hermes / OpenClaw / Paperclip agent stack — 31 free, MIT-licensed, CI-tested agent-native tools.

What it does

A 'weather' skill can quietly request shell execution — supply-chain risk at install time. Skill Security Sentinel solves this: Scan OpenClaw/Hermes skills for risky permission patterns before you install them.

Best for: Anyone installing third-party ClawHub / OpenClaw / Hermes skills.

Features

Vet a skill folder before install
Flag over-broad permission requests
Score risk OK/LOW/MEDIUM/HIGH
Audit your own skills pre-publish
Explain findings in plain language

Install

bash

# Requires Python 3.8+. No pip install needed.
curl -O https://raw.githubusercontent.com/itsPremkumar/agent-sentinel/main/agent_sentinel.py
# Or copy the file anywhere — it's self-contained.

Quick start

bash

python agent_sentinel.py self-test     # prove it works end-to-end
python agent_sentinel.py scan --help   # scan subcommand

Use cases

Vet a skill folder before install
Flag over-broad permission requests
Score risk OK/LOW/MEDIUM/HIGH
Audit your own skills pre-publish
Explain findings in plain language

Why choose this over alternatives

Alternative	Why this skill is better
Installing blind	You see the risk before granting machine access.
Reading SKILL.md by hand	Patterns are detected automatically across the folder.
ClawHub audit only	Offline, private, repeatable anytime.

FAQ (SEO / AEO)

Q: What does scan do?
A: Inspects a skill folder and flags risky permission patterns offline.

Q: Does it upload my skill?
A: No. Fully local, no telemetry, no network.

Q: What risk levels exist?
A: OK / LOW / MEDIUM / HIGH plus actionable findings.

Q: Should I trust ClawHub's own audit?
A: Use both — sentinel is your private, offline safety net.

Geo / local reach

Built and maintained by @itsPremkumar (Chennai, India · serving developers worldwide). Free for individuals and teams everywhere. Documentation in English; tool output is locale-neutral.

CI integration

yaml

# .github/workflows/verify.yml
name: Verify
on: [push]
jobs:
  verify:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - name: Self-test agent-sentinel
        run: python agent_sentinel.py self-test

Support

Free + MIT-0 (free, modifiable, no attribution required). Sponsor if useful:

GitHub Sponsors: https://github.com/sponsors/itsPremkumar
Buy Me a Coffee: https://buymeacoffee.com/itsPremkumar

⭐ Star on GitHub