skill-test
Pass
Audited by Gen Agent Trust Hub on Jul 10, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill acts as an automated auditor and linter, reading and processing the contents of external skill files and test specifications. This introduces a surface for indirect prompt injection.\n
- Ingestion points: Markdown content from local skill files (.agents/skills/*/SKILL.md) and YAML data from catalog and spec files.\n
- Boundary markers: There are no instructions or delimiters defined to isolate untrusted skill content from the auditing instructions.\n
- Capability inventory: The skill utilizes Read, Glob, Grep, and Write tools for file inspection and recording results.\n
- Sanitization: No sanitization or validation of the ingested skill text is performed before the agent evaluates its behavioral assertions.
Audit Metadata