ts-type-audit
Pass
Audited by Gen Agent Trust Hub on Nov 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data by reading and analyzing project source code, which creates a surface for indirect prompt injection if the audited files contain adversarial instructions.
- Ingestion points: The skill reads all TypeScript source files within a user-specified directory and sub-perimeter (e.g.,
src/,scripts/) to perform the audit. - Boundary markers: The orchestration logic in
SKILL.mduses clear Markdown delimiters (e.g.,--- Référence du tier ---,--- Rapport... ---) to separate instruction blocks, which helps but does not fully eliminate injection risks. - Capability inventory: The skill utilizes file reading and search capabilities (
grep) to ingest code and produces structured reports in a designated output directory (reports/ts-type-audit/). - Sanitization: The instructions explicitly direct the worker to exclude code contained within strings (fixtures, lint rule counter-examples) from the analysis, reducing the likelihood of accidental instruction following from data literals.
Audit Metadata