agent-collective-intelligence-coordinator
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external agents to build a collective state, which creates a vulnerability to instructions embedded in that data.
- Ingestion points: The skill explicitly aggregates inputs from other agents and specialized sub-agents (referenced in 'Consensus Building' and 'Cognitive Load Balancing' sections in SKILL.md).
- Boundary markers: Absent. There are no instructions to use delimiters or specific ignore-tags when processing aggregated agent inputs.
- Capability inventory: The skill uses the
mcp__claude-flow__memory_usagetool to store and synchronize states, meaning injected instructions could persist in the shared memory context. - Sanitization: Absent. The skill does not provide logic for validating or filtering external inputs before integrating them into the hive consensus.
Audit Metadata