architect
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [SAFE]: No malicious patterns, prompt injections, or unauthorized data access were found.
- [NO_CODE]: The skill does not include any executable scripts or binaries, consisting entirely of markdown instructions and templates.
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by processing external repository data. 1. Ingestion points: Reads AGENTS.md, repository code, dependency manifests, and configuration files as specified in SKILL.md. 2. Boundary markers: Absent. The skill does not implement delimiters to isolate instructions from ingested file content. 3. Capability inventory: The skill is limited to reading repository files and writing markdown documentation (architecture, requirements, ADRs) and metadata files (AGENTS.md, README.md). 4. Sanitization: Absent. There is no mention of sanitizing or escaping ingested data. Assessment: The risk is safe as the skill's capabilities are restricted to documentation generation and do not include network or shell execution tools.
Audit Metadata