action-pattern

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill directs the agent to analyze the existing codebase and repository structure to design and implement new business logic (Actions). This processing of external, potentially untrusted source code introduces a surface for indirect prompt injection attacks where malicious instructions could be embedded in the repository content.
  • Ingestion points: The agent is instructed to 'inspect the repository' for existing conventions, dependencies, and rules in the 'Required Inputs And Prerequisites' section.
  • Boundary markers: The skill lacks explicit delimiters or instructions to treat repository code as untrusted data or to ignore embedded instructions.
  • Capability inventory: The skill grants the agent the ability to implement and modify source files ('Implement the Action in its own source file') and update entry points like controllers or jobs.
  • Sanitization: No validation or sanitization mechanisms for the ingested source code content are defined in the instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 06:45 AM
Security Audit — agent-trust-hub — action-pattern