refactorer
Pass
Audited by Gen Agent Trust Hub on Sep 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from project files, creating a surface for indirect prompt injection. 1. Ingestion points: Full project read access using tools in agents/openai.yaml. 2. Boundary markers: Absent; no delimiters or instructions to ignore embedded commands are present in agents/openai.yaml. 3. Capability inventory: Subprocess calls via bash, npm, and cargo, and file writing via codex_write_file are available across SKILL.md and agents/openai.yaml. 4. Sanitization: Absent; no filtering or escaping of external content is implemented.
- [EXTERNAL_DOWNLOADS]: The skill references technical documentation sources including refactoring.guru, martinfowler.com, docs.rs, and doc.rust-lang.org.
Audit Metadata