agent-architect

Pass

Audited by Gen Agent Trust Hub on Jun 21, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted content from the target project's codebase to generate architectural profiles and recommendations, which creates a potential surface for indirect prompt injection.
  • Ingestion points: Reads codebase manifests (e.g., package.json, pyproject.toml), directory listings, and existing AI configurations (CLAUDE.md, AGENTS.md) as specified in SKILL.md and references/codebase-signals.md.
  • Boundary markers: The instructions do not define explicit delimiters or 'ignore embedded instructions' warnings for the codebase content being read.
  • Capability inventory: The skill has shell, filesystemRead, and filesystemWrite capabilities enabled according to skill.json.
  • Sanitization: There are no explicit sanitization or validation steps for the content ingested from the codebase before processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 21, 2026, 05:49 AM
Security Audit — agent-trust-hub — agent-architect