agent-collective-intelligence-coordinator

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION] (LOW): Indirect Prompt Injection Surface.
  • Ingestion points: The skill explicitly aggregates inputs from multiple external sources including 'all agents' in the hive and specific agents like 'scout-explorer' (File: SKILL.md).
  • Boundary markers: Absent. There are no instructions to use delimiters or 'ignore embedded instructions' warnings when processing these distributed inputs.
  • Capability inventory: The skill utilizes the 'mcp__claude-flow__memory_usage' tool to write to shared memory keys (e.g., 'swarm$shared$collective-state'), which are intended to be read by other high-priority agents.
  • Sanitization: Absent. No logic is provided to escape, validate, or filter content gathered from the environment before it is committed to the shared hive memory.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:11 PM