skills/product-on-purpose/thinking-framework-skills/think-interest-based-negotiation/Gen Agent Trust Hub
think-interest-based-negotiation
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [SAFE]: No malicious patterns, obfuscation, or security vulnerabilities were identified in the skill instructions or supporting files.
- [SAFE]: The skill is designed for preparation deskwork only and does not contain any instructions or tools for performing network operations, file system modifications, or command execution.
- [SAFE]: The framework incorporates safety-oriented instructions regarding information disclosure, specifically requiring the agent to flag which interests are safe to reveal and which should be held.
- [INDIRECT_PROMPT_INJECTION]: The skill includes an attack surface for indirect prompt injection. 1. Ingestion points: Negotiation details are provided by the user in SKILL.md. 2. Boundary markers: Present via the structured template in references/TEMPLATE.md. 3. Capability inventory: No subprocess calls, file-writes, or network operations are present in any script or instruction. 4. Sanitization: Absent. The risk is assessed as safe due to the complete lack of exploitable capabilities.
- [EXTERNAL_DOWNLOADS]: The skill references documentation and framework metadata from the author's official GitHub repository (product-on-purpose/thinking-framework-skills).
Audit Metadata