architecture-review

Pass

Audited by Gen Agent Trust Hub on Nov 11, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external content, including architecture specifications, RFCs, and source code, which could contain malicious instructions intended to manipulate the agent's review logic or findings.
  • Ingestion points: Reads files such as REQUIREMENTS.md, ARCHITECTURE.md, and any provided technical proposals or codebases.
  • Boundary markers: The instructions do not define clear delimiters or use prompt engineering techniques to prevent the agent from obeying instructions embedded within the data being reviewed.
  • Capability inventory: The skill requires file-read access to project documents and suggests the execution of command-line tools for sub-agent reviews.
  • Sanitization: There is no evidence of sanitization, filtering, or validation for the content of the analyzed documents.
  • [COMMAND_EXECUTION]: The instructions explicitly direct the agent to invoke subagents using shell-based commands such as codex exec "<prompt>" or claude -p "<prompt>" to perform read-only reviews.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 11, 2026, 05:04 AM