mcp
Pass
Audited by Gen Agent Trust Hub on Nov 12, 2026
Risk Level: SAFEREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill documents dangerous code patterns, such as string interpolation in shell commands, strictly for educational purposes to demonstrate vulnerabilities and teach the use of secure alternatives like argument arrays.
- [EXTERNAL_DOWNLOADS]: The skill references and fetches developer tools from the official Model Context Protocol repository on GitHub, which is a well-known service related to the skill's purpose.
- [DATA_EXFILTRATION]: References to sensitive system paths (e.g.,
~/.ssh/id_rsaand/etc/passwd) are used exclusively within documentation describing tool poisoning attacks and security test suites. - [INDIRECT_PROMPT_INJECTION]: The skill includes detailed guidance on defending against malicious instructions embedded in tool descriptions or dynamic metadata, identifying this as a potential attack surface and providing specific defensive strategies.
Audit Metadata