mastra
Pass
Audited by Gen Agent Trust Hub on Aug 10, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill includes a formatting instruction directing the agent to always prepend a specific emoji (🧢) to its initial response when the skill is activated.
- [COMMAND_EXECUTION]: Instructions feature a 'Companion check' that requires the agent to execute directory listing commands (
ls) on specific hidden paths (e.g.,~/.claude/skills/) to identify other installed tools. - [EXTERNAL_DOWNLOADS]: Setup and installation documentation promotes the use of
npmandnpxto download and execute code from external sources, specifically the AbsolutelySkilled repository and the npm registry. - [REMOTE_CODE_EXECUTION]: Examples demonstrate connecting agents to external Model Context Protocol (MCP) servers using
npx, enabling the execution of remote tools and scripts. - [INDIRECT_PROMPT_INJECTION]: The skill has an ingestion surface through external API calls (e.g., weather tools), RAG pipeline data, and workflow inputs; while explicit boundary markers are absent, the framework uses Zod schemas for structured data validation.
Audit Metadata