Loading…
Scan MCP servers & AI-agent skills for malicious patterns (prompt injection in tool descriptions, credential exfil, download-and-execute). 17 patterns / 60 signatures, zero-dep. Measured over 196 public MCP servers: 99% clean, false-positive rate calibrated 14.8% -> 1.0%.
mcp-audit is an MCP server entry in ChatHome Skills, sourced from GitHub MCP Search and scored 26/100. Scan MCP servers & AI-agent skills for malicious patterns (prompt injection in tool descriptions, credential exfil, download-and-execute). 17 patterns / 60 sign. As a Model Context Protocol server, it exposes tools that MCP-compatible LLM clients such as Claude, Cursor, and Copilot can call directly, extending an agent's reach beyond its built-in capabilities. ChatHome Skills discovered this entry through its daily ingest of official registries and GitHub discovery queries, and re-verifies repository facts, freshness signals, and install metadata on every run. The quality score weighs official-registry presence, a working install command, license clarity, GitHub star trajectory, and repository freshness — the full formula is on the Sources page. The install command above is ready to copy. Review the last-pushed date and repository status before adding mcp-audit to a production agent workflow; the snapshot facts below are point-in-time values from the most recent ingest.
git clone https://github.com/eltociear/mcp-audit.gitReview the upstream source before running this command.
Downloads a script and pipes it straight into a shell, executing remote code with no review.
curl … \| shDispute this flag?Contains language that tries to override an agent's prior instructions (prompt injection).
ignore previous instructionsDispute this flag?Scan coverage: Install metadata and repository README.
Automated signals from install metadata and a bounded skill body — not a full code security audit and not a recursive source or dependency review.
{
"fullName": "eltociear/mcp-audit"
}Scan MCP servers & AI-agent skills for malicious patterns (prompt injection in tool descriptions, credential exfil, download-and-execute). 17 patterns / 60 signatures, zero-dep. Measured over 196 public MCP servers: 99% clean, false-positive rate calibrated 14.8% -> 1.0%.