The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The installation process involves downloading a package from an unverified community repository on GitHub (dbhat93/job-search-os).
[COMMAND_EXECUTION]: The skill documentation provides an npx command for installation, which fetches and executes arbitrary code from a remote, unverified source.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection through its transcript analysis and storybank features.
Ingestion points: The skill ingests untrusted data when users paste interview transcripts from external tools like Otter, Zoom, or Grain (SKILL.md, Example 3).
Boundary markers: There are no specified boundary markers or instructions to the agent to disregard instructions embedded within the processed transcript data.
Capability inventory: The skill performs file system operations, specifically reading and writing to coaching_state.md to persist user data and coaching state.
Sanitization: No sanitization or validation mechanisms are described for the external content before it is processed by the agent.

interview-coach