cyber-intel

Pass

Audited by Gen Agent Trust Hub on Aug 27, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill provides a structured set of guidelines for an 'Agentic Company Framework,' defining operational modes such as BUILD, INCIDENT, and EXPERIMENT to govern agent behavior during development tasks.
  • [COMMAND_EXECUTION]: Instructions specify the use of an 'ExecutionProxy Interface' (using an rtk prefix, e.g., rtk npm test) to handle terminal actions. This is presented as a method for deterministic execution and computational efficiency rather than a mechanism for unauthorized command injection.
  • [INDIRECT_PROMPT_INJECTION]: The protocols explicitly address the handling of untrusted inputs, mandating that web content and external data be treated as hostile and requiring the redaction of secrets/PII before context sharing. This represents a defensive security implementation to prevent injection attacks.
  • [DATA_EXFILTRATION]: The skill mentions accessing local project files like package.json and docs/graph.json for the purposes of version pinning and architecture discovery, which are standard operations for development tools and do not involve unauthorized data movement.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 27, 2026, 07:35 AM
Security Audit — agent-trust-hub — cyber-intel