architect
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-supplied design tasks and architectural grounding artifacts to generate code sketches and implementations, which is a vector for indirect prompt injection.\n
- Ingestion points: External design tasks and grounding artifacts are ingested in
SKILL.md(Phase B: Sketch) and passed to candidate runners.\n - Boundary markers: Absent; the skill does not use delimiters or instructions to ignore embedded commands in the design input.\n
- Capability inventory: The skill can spawn subagents (Phase B) and modify codebase files by replacing placeholders with generated code (Phase D) in
SKILL.md.\n - Sanitization: Absent; requirements are interpolated into prompts without validation or escaping.
Audit Metadata