AI agent security scanner that detects vulnerabilities in agent configurations, MCP servers, and tool permissions.
agentshield is easy to set up with strong trust signals. Check agent compatibility and use-case fit before adding it to your workflow.
Repository setup guidance
gh repo view affaan-m/agentshield --webOpen the official repository or website.
Check the README for package manager, auth, and platform requirements.
Try it in a small test task inside your agent workflow.
AgentShield scans your AI agent setup (like Claude Code) for security issues. It finds hardcoded secrets, dangerous permissions, and other misconfigurations before attackers can exploit them. You can run it as a command-line tool, GitHub Action, or GitHub App.
AgentShield is a security auditor for AI agent configurations, specifically designed for Claude Code setups. It scans your `.claude/` directory for hardcoded secrets, permission misconfigurations, hook injection vulnerabilities, MCP server risks, and agent prompt injection vectors. The tool auto-discovers your configuration, scans all relevant files, and produces a graded security report (A-F). It supports multiple output formats (CLI, JSON, HTML) and can auto-fix safe issues by replacing hardcoded secrets with environment variable references. AgentShield also includes an optional Opus 4.6 adversarial analysis mode that uses three AI agents to simulate attacks. It is part of the Everything Claude Code ecosystem and was built at the Claude Code Hackathon by Cerebral Valley and Anthropic.
Strong trust signals; still review the README and permissions before production use.
Last commit was about 52 days ago.
813 GitHub stars indicate community interest.
2 open issues signal maintenance load.
MIT license detected.
Scan Claude Code configurations for hardcoded API keys and secrets before deployment
Audit MCP server configurations for security risks and misconfigurations
Integrate into CI/CD pipelines via GitHub Action to automatically block insecure agent setups
Generate executive security reports in HTML format for compliance reviews
Auto-fix common security issues like replacing hardcoded secrets with environment variables
The tool requires access to your Claude Code configuration files, which may contain sensitive data. Use with caution in shared environments.
Auto-fix mode modifies configuration files; ensure backups are taken before running with --fix flag.
Opus analysis mode sends configuration data to Anthropic's API for adversarial testing; review data handling policies.
813
Stars
173
Forks
2
Issues
MIT
License
A terminal dashboard to monitor AI coding agents like Claude Code and Codex CLI in real-time.
A bridge between Streamable HTTP and stdio MCP transports, enabling flexible MCP server connectivity.
A full-stack AI Red Teaming platform for securing AI ecosystems with comprehensive scanning and evaluation.
AI agent security scanner that detects vulnerabilities in agent configurations, MCP servers, and tool permissions.
Security engineers auditing AI agent deployments, Developers using Claude Code who want to secure their configurations, DevOps engineers integrating agent security into CI/CD pipelines
The setup section provides repository-level starting guidance, not a guarantee of an independently verified installation. Check the official README and release notes.
The tool requires access to your Claude Code configuration files, which may contain sensitive data. Use with caution in shared environments. Auto-fix mode modifies configuration files; ensure backups are taken before running with --fix flag.
Similar or complementary options to evaluate include abtop, mcp-proxy, AI-Infra-Guard.
Jun 6, 2026
3 security/trust notes recorded.
Setup difficulty is 1/5.