influence-and-negotiation

Pass

Audited by Gen Agent Trust Hub on Apr 30, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill's workflow involves ingesting and analyzing untrusted data from multiple external channels, creating a surface for indirect prompt injection.
  • Ingestion points: In references/context-intake.md (Step 1 and Step 2), the agent is instructed to collect raw material from potentially attacker-controlled sources, including email threads, Slack messages, LinkedIn profiles, and general web search results.
  • Boundary markers: There are no explicit instructions or structural delimiters (e.g., specific tags or 'ignore' commands) to ensure the agent disregards instructions that might be embedded in the processed external content.
  • Capability inventory: The skill possesses high-privilege capabilities, including file system access (Read, Edit, Write) and the ability to delegate tasks to sub-agents (Agent tool). Malicious instructions embedded in external data could potentially abuse these tools.
  • Sanitization: The provided instructions do not include steps for validating, filtering, or escaping content retrieved from external sources before it is processed or stored in the local memory files.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 30, 2026, 02:33 AM