coach
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill features a dedicated 'External Text Validation Module' designed to scan untrusted inputs such as transcripts and job descriptions for embedded instructions (prompt injection), explicitly instructing the agent to flag these and seek user confirmation.
- [SAFE]: Data privacy is handled appropriately; sensitive professional and compensation data is stored in local markdown files that are managed with explicit instructions to be added to .gitignore to prevent cloud synchronization or exposure.
- [SAFE]: The use of dynamic context injection via shell commands (e.g., 'cat' to load instruction files or 'date' for accuracy) is restricted to utility tasks and does not process unvalidated user input.
- [SAFE]: All external research is conducted via established tools (Exa MCP) and results are rigorously labeled with confidence levels and source tiers to prevent the generation of fabricated information.
Audit Metadata