bfs-prompt-architect

Pass

Audited by Gen Agent Trust Hub on Jul 14, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted user-provided content (referred to as a "rough brief") to synthesize technical specifications and implementation prompts. This architectural pattern is susceptible to indirect prompt injection, where a malicious brief could contain instructions intended to bypass safety filters or manipulate the agent's final output.
  • Ingestion points: The user input brief is ingested via the argument-hint parameter in SKILL.md.
  • Boundary markers: The skill does not define clear delimiters or "ignore instructions" warnings to isolate user input from its own internal logic.
  • Capability inventory: The skill is restricted via disable-model-invocation: true and lacks capabilities for file writing, network access, or subprocess execution.
  • Sanitization: No validation or sanitization of the user-provided brief is performed prior to its transformation into a product spec.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 14, 2026, 03:44 PM
Security Audit — agent-trust-hub — bfs-prompt-architect