architect
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection by ingesting data from external subsystems during the grounding phase.
- Ingestion points: In Phase A ("Ground the problem"), the instructions require running the
howskill over relevant subsystems and files to build a mental model of the target environment. - Boundary markers: The skill does not specify the use of delimiters, XML tags, or "ignore embedded instructions" warnings to isolate grounding artifacts from the primary agent instructions.
- Capability inventory: The agent possesses significant capabilities including the ability to write new source code (Phase D), modify files, and invoke other functional skills like
arenaandinterrogate. - Sanitization: There is no mention of sanitizing, filtering, or escaping codebase content before it is processed by the model during architectural synthesis.
Audit Metadata