The Agent Skills Directory

[SAFE]: The skill provides structured coaching for job searches and interviews. It implements session persistence by storing data in a local file, coaching_state.md, which is a standard method for maintaining context in agentic workflows without requiring external databases or elevated privileges.
[EXTERNAL_DOWNLOADS]: The skill is installed via a standard command, npx skills add dbhat93/job-search-os, and points to a repository on GitHub. These are well-recognized platforms for code distribution, and no suspicious or unverified download patterns were detected.
[PROMPT_INJECTION]: The skill is designed to process external content such as job descriptions and interview transcripts, which constitutes a potential surface for indirect prompt injection.
Ingestion points: Inputs processed through commands like JD decoding and transcript analysis (SKILL.md).
Boundary markers: None identified in the provided instructions.
Capability inventory: The skill is limited to the claude tool; it does not have access to shell execution, sensitive system files, or arbitrary network operations.
Sanitization: No specific sanitization logic is documented. Due to the limited capabilities and the primary purpose of the skill, this surface is considered a standard functional risk rather than a malicious finding.

interview-coach