tech-debt-audit
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection vulnerability surface when auditing external codebases.
- Ingestion points: Ingests untrusted data by scanning repository files (
**/*.ts,**/*.py,package.json,README.md) usingread(),grep, and AST-grep (sg) inSKILL.mdacross multiple audit dimensions. - Boundary markers: Lacks boundary markers or safety instructions instructing the agent or sub-agents to disregard embedded prompt injections within code comments, strings, or documentation.
- Capability inventory: Possesses high-privilege capabilities including arbitrary terminal execution via
bash()and background sub-agent creation viatask()blocks in Phase 2. - Sanitization: No input sanitization, filtering, or escaping is performed on the files read before processing them or interpolating data into sub-agent contexts.
Audit Metadata