cv-scorer
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted third-party inputs (candidate CVs/resumes) which may contain hidden prompt injection vectors aimed at inflating scores or altering agent behavior.
- Ingestion points: Identified in
SKILL.md(Step 1: Identify Inputs), where the agent reads CV documents provided in text, markdown, or PDF formats. - Boundary markers: Absent. The instructions do not specify using strict xml tags, markdown blocks, or delimiters to isolate untrusted data, nor do they instruct the model to ignore any systemic or behavior-altering text found within the CV.
- Capability inventory: There are no dangerous execution capabilities such as code execution (
eval/exec), system commands, file writing, or external network activities. The risk is limited to output text manipulation and incorrect structured JSON reporting. - Sanitization: Absent. No predefined formatting validations or structural verification guidelines are implemented for processing external documents.
Audit Metadata