algo-net-influence
Pass
Audited by Gen Agent Trust Hub on Apr 10, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection through the ingestion of external network graph data. Ingestion points: Network data (nodes and edges) is ingested in SKILL.md under 'Phase 1: Input Validation'. Boundary markers: The instructions lack delimiters or explicit warnings to the agent to ignore instructions embedded within the graph data. Capability inventory: The skill provides Python implementation code in references/celf-implementation.md and references/scalable-im.md for the agent to execute for simulations. Sanitization: There is no evidence of sanitization or content validation to prevent malicious strings in the graph from influencing the agent's behavior.
Audit Metadata