affaan-maffaan-mCommunity·1 Agent Skills Included

agentshield

Audit AI agent configs for secrets, permissions, and injection risks

Scans Claude Code setups, MCP servers, hooks, and agent configs for hardcoded secrets, wildcard permissions, and prompt injection vectors. Applies 102 security rules across five categories and grades your setup A to F with a 0-100 score. Auto-fixes safe issues, verifies supply-chain provenance, and exports JSON, HTML, or SARIF reports for CI pipelines.
npx skills add affaan-m/agentshield --all -g -y
Available:

Gives the AI agent the project architecture, rule modules, severity scoring model, and CLI commands so it can navigate, extend, and test the AgentShield codebase correctly.

All Skills in This Repository (1)

Pure Emerald Level Indicators

Frequently Asked Questions

FAQPage Schema
How to install AgentShield?

Run `npx skills add affaan-m/agentshield --all -g -y` in your terminal to install the full suite globally.

What does AgentShield scan for?

It checks your Claude Code and MCP configuration for hardcoded API keys, overly broad permissions like Bash(*), hook injection, supply-chain risks, and prompt injection vectors across 102 rules.

Can AgentShield fix security issues automatically?

Yes. Running a scan with the fix flag replaces hardcoded secrets with environment variable references and suggests safe permission restrictions.

Does AgentShield work in CI pipelines?

Yes. It ships as a CLI, GitHub Action, and GitHub App, and outputs JSON or SARIF reports so builds can fail on critical findings.

Do I need security expertise to use AgentShield?

No. Run one scan command and it prints a graded A-F report with plain-English explanations and fix suggestions for every finding.

Related Repositories in Software Engineering

View All in Software Engineering