patent-pipeline

Pass

Audited by Gen Agent Trust Hub on Apr 19, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it ingests untrusted data from external web searches (via /prior-art-search) and processes user-provided invention descriptions. These inputs are interpolated into prompts for subsequent drafting and review phases without explicit sanitization or strict boundary markers.
  • Ingestion points: WebSearch results, user-provided INVENTION_BRIEF.md, and conversational inputs from $ARGUMENTS.
  • Capability inventory: The skill has access to file-writing tools, shell execution, and the ability to trigger other agent skills.
  • Boundary markers: None explicitly defined in the provided instruction set to isolate external content from the agent's core logic.
  • [COMMAND_EXECUTION]: The skill uses the Bash tool for file operations and state management. This is a standard use case for the ARIS architecture and is used to handle large file writes via heredocs, which is a legitimate practice in this environment.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 19, 2026, 03:14 AM