mcpsec-skill_skill

This skill scans MCP server configuration files for security issues using mcpsec, reporting findings by severity without altering any files.
  • Python

2.5k

GitHub Stars

3

Bundled Files

2 months ago

Catalog Refreshed

3 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill openclaw/skills --skill mcpsec-skill

  • _meta.json454 B
  • README.md2.6 KB
  • SKILL.md5.9 KB

Overview

This skill scans Model Context Protocol (MCP) server configuration files for security vulnerabilities using the open-source mcpsec scanner. It auto-discovers common config locations (Claude Desktop, Cursor, VS Code, custom paths), reports findings by severity, and runs read-only — it never modifies configs. Use it to audit configs for OWASP MCP Top 10 issues like prompt injection, hardcoded secrets, missing auth, insecure transport, and excessive permissions.

How this skill works

The Python wrapper validates user-supplied paths against an allowlist, then invokes the mcpsec binary (subprocess with shell=False) to analyze discovered JSON config files. mcpsec checks coverage across the OWASP MCP Top 10 and returns findings grouped by severity; the wrapper supports JSON output, severity filtering, and a quiet mode. The wrapper itself performs no network calls and never writes or transmits configuration files.

When to use it

  • Performing a security audit of local MCP tool configurations
  • Checking for hardcoded API keys or secrets in config files
  • Validating that tools have appropriate permissions and authentication
  • Detecting insecure transport settings (HTTP/weak TLS) or SSRF risks
  • Running scheduled checks (cron) with quiet mode to report only when issues exist

Best practices

  • Verify provenance and checksums of the mcpsec binary before installing
  • Run scans in an isolated environment or container if you require stronger isolation
  • Use severity filters to focus on critical/high findings in automation
  • Use JSON output for integration with dashboards or SIEM systems
  • Review findings before remediating; configs can contain sensitive tokens — treat output as sensitive

Example use cases

  • One-off audit of Claude Desktop and VS Code MCP configs before deploying an assistant
  • CI job that scans committed MCP config files and fails on critical/high findings
  • Periodic cron job that runs --quiet and alerts only when new vulnerabilities are found
  • Security triage: export JSON findings to a ticketing system for remediation tracking

FAQ

No. The wrapper and mcpsec operate in read-only mode; no files are modified by this skill.

Does the scanner send my config data over the network?

The wrapper makes no network calls. The upstream mcpsec binary is designed to read local files only, but you should verify the binary and checksums if you need absolute assurance.

Built by
VeilStrat
AI signals for GTM teams
© 2026 VeilStrat. All rights reserved.All systems operational