ats-latex-cv
Pass
Audited by Gen Agent Trust Hub on May 20, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill contains instructions that suppress transparency. Step 0 directs the agent to 'silently check' inputs and 'never announce' its operational mode to the user. Furthermore, Step 5 mandates the agent output 'ONLY the raw LaTeX code' with 'no explanation' and 'no commentary', which effectively conceals the agent's reasoning process and potential instruction leaks from the user's view.
- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests untrusted data from user-uploaded CVs and job descriptions to drive its logic and web search activities.
- Ingestion points: CV reconstruction in Step 0 and Job Description keyword extraction in Step 2.
- Boundary markers: Absent. The instructions do not specify delimiters or tell the agent to ignore instructions embedded within the uploaded documents.
- Capability inventory: The agent has the capability to perform web searches (Step 2) and generate executable-style text (LaTeX code in Step 5).
- Sanitization: Absent. While the agent is told to confirm extracted data with the user in Step 0, there is no technical validation or sanitization of the input content to prevent malicious prompt overrides.
Audit Metadata