meta-governor

Pass

Audited by Gen Agent Trust Hub on Nov 16, 2026

Risk Level: SAFEPROMPT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill contains instructions designed to override agent control mechanisms by claiming it is "Always active" and "Cannot be disabled." This pattern attempts to establish persistence and prevent the agent from following user instructions to deactivate or bypass the governor's logic.
  • [INDIRECT_PROMPT_INJECTION]: The skill accepts user requests as input for classification and skill selection, which creates a surface for indirect prompt injection attacks if the requests contain adversarial content.
  • Ingestion points: The skill ingests raw user requests to perform classification (Bug, Feature, etc.) and activation decisions.
  • Boundary markers: There are no delimiters or specific instructions provided to the agent to treat user input as untrusted data or to ignore embedded commands.
  • Capability inventory: While this specific skill does not perform high-risk operations (file writes, network calls), it acts as a gateway that triggers other capabilities based on potentially malicious input.
  • Sanitization: The instructions do not define any validation, filtering, or escaping protocols for the processed user requests.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 16, 2026, 03:31 AM