code-roaster
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted data (source code files) without adequate isolation or boundary markers.
- Ingestion points: The skill uses the Glob tool to read various code file types (JS, TS, Python, Java, etc.) as specified in
SKILL.md. - Boundary markers: The instructions do not specify the use of delimiters or 'ignore embedded instructions' warnings when the agent reads the source code for analysis.
- Capability inventory: The skill has the ability to read files, search the filesystem via Glob, and write output to a new file (
CODE_ROAST_REPORT.md). - Sanitization: There is no evidence of sanitization or filtering of the content read from external files before it is processed by the agent. A malicious instruction embedded in a comment within a scanned code file could potentially influence the agent's behavior during the 'roasting' process.
Audit Metadata