resume-tailoring
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external job descriptions and web-fetched profiles (from LinkedIn and GitHub) to synthesize a success profile. This creates an attack surface for indirect prompt injection where malicious instructions embedded in these external sources could attempt to influence the agent's logic.
- Ingestion points: The skill processes Job Descriptions provided as text or URLs in Phase 1 and analyzes LinkedIn/GitHub profiles in Phase 1/benchmarking.
- Boundary markers: The prompts in
research-prompts.mdfor parsing and benchmarking do not utilize explicit delimiters or "ignore" instructions for the ingested content. - Capability inventory: The skill can write files to the local filesystem (
resumes/directory) and perform network operations usingWebSearchandWebFetchtools. - Sanitization: No evidence of input sanitization or filtering for the retrieved web content was found.
- [EXTERNAL_DOWNLOADS]: The skill performs network operations using
WebSearchandWebFetchto retrieve data from various internet sources, including company engineering blogs, news sites, and professional profiles on well-known platforms like LinkedIn and GitHub for role benchmarking. - [COMMAND_EXECUTION]: The documentation provides shell commands (e.g.,
npx,cp,mkdir) as instructions for users to manually install and set up the skill in different agent environments.
Audit Metadata