design-code-architecture
Pass
Audited by Gen Agent Trust Hub on Aug 23, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill recommends that users install external sub-skills from an unverified repository (wondelai/skills) using the npx skills add command if a required dependency is missing.
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through its session resumption mechanism, which reads untrusted data from project files to determine its state and next steps. 1. Ingestion points: Where untrusted data enters agent context (docs/DESIGN-CODE-ARCHITECTURE-PLAN.md and docs/ARCHITECTURE.md). 2. Boundary markers: Delimiters or 'ignore embedded instructions' warnings (Absent). 3. Capability inventory: All subprocess calls, exec/eval, file-write, network ops (Invokes sub-skills, writes/modifies documentation files). 4. Sanitization: Escaping, validation, or filtering of external content (Absent).
- [PROMPT_INJECTION]: There is a metadata author mismatch between the provided context (manish31-oss) and the skill frontmatter (wondelai), which could be misleading to users or other agents regarding the skill's origin.
Audit Metadata