garagon

garagon / aguara

Public

Security scanner for AI agent skills & MCP servers. 173 detection rules. 13 categories. 5 registries monitored daily. OpenClaw detection included. No API keys, no cloud, no LLM. One binary. Detection engine behind Oktsec.

aguarascan.com ai-agents claude claude-code golang mcp
45
5
100% credibility
Found Feb 20, 2026 at 16 stars 3x -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Go
AI Summary

Aguara scans AI agent skill files and server setups for security risks like prompt tricks, data leaks, and sneaky attacks.

How It Works

1
🔍 Discover safety checker

You hear about Aguara while reading about keeping AI helper files safe from hidden tricks.

2
📥 Get the tool

Download the simple safety checker app that works on your computer.

3
📁 Pick your folder

Choose the folder holding your AI skill descriptions, like helpful guides for your assistants.

4
🚀 Run the safety scan

Click to scan and it quickly checks everything for sneaky problems like data leaks or bad instructions.

5
📋 See the results

Get a clear report highlighting any issues with colors and easy explanations.

✅ Skills are secure

Your AI helpers pass the check and you're ready to use them confidently without worries.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 16 to 45 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is aguara?

Aguara is a Go-based static security scanner for AI agent skills and MCP servers, delivering one binary that runs locally with no API keys, cloud, or LLMs. It applies 138+ rules across 15 categories to detect prompt injections, data exfiltrations, credential leaks, and supply-chain attacks in markdown skills directories like .claude/skills/. Output JSON, SARIF, or Markdown for github security scanning in CI, with deterministic results and flags like --ci or --changed for github actions.

Why is it gaining traction?

Unlike regex-only tools or LLM-dependent scanners, Aguara handles obfuscated attacks via markdown-aware analysis and decodes base64/hex payloads, while staying reproducible for CI fail thresholds. Extensible YAML rules let you add custom checks without recompiling, and SARIF integration feeds directly into github security alerts and code scanning. Developers grab it for quick, private scans that catch AI-specific threats missed by general github security advisories.

Who should use this?

AI agent builders maintaining Claude skills or MCP servers, especially teams deploying via github actions needing security github copilot safeguards. Security engineers scanning repos with untrusted third-party AI tools, or ops folks gating PRs on high/critical findings in skill markdowns.

Verdict

Try Aguara if you're in AI agents—solid docs, full test coverage, and CI-ready despite 11 stars and 1.0% credibility signaling early maturity. Pairs well with security github projects for a free security scanner alternative.

(178 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.