ptlam-architecturing

Pass

Audited by Gen Agent Trust Hub on Sep 2, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and analyze external architectural documents, ADRs, and domain-specific terms, which creates an attack surface for indirect prompt injection if those documents contain malicious instructions.
  • Ingestion points: In SKILL.md under '1. Start from one question', the agent is directed to read existing architecture documents, Architecture Decision Records (ADRs), and domain terms from CONTEXT.md.
  • Boundary markers: The instructions do not mandate the use of delimiters (e.g., XML tags or triple quotes) or specific 'ignore embedded instructions' warnings when processing these external files.
  • Capability inventory: The skill is primarily instructional but allows for file writing to user-named destinations and the generation of visual diagrams using the ptlam-mermaiding sub-skill.
  • Sanitization: There are no explicit steps provided for sanitizing, escaping, or filtering the content of external documents before they are interpolated into the agent's context.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 2, 2026, 09:47 AM
Security Audit — agent-trust-hub — ptlam-architecturing