never-split-the-difference-skill
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious patterns, obfuscation, or unauthorized data access were identified during the analysis of the instruction set and configuration files.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided negotiation scenarios (e.g., deal details, prospect objections) to generate strategies and drafts. While this involves processing untrusted data, the risk is addressed through the application of the
check_my_copybrand validation tool and the use of rigid negotiation templates that limit the impact of potentially malicious input. - [DATA_EXPOSURE]: The skill records operational metrics and run logs to specific local files (e.g.,
~/.claude/skill-analytics/). This activity is restricted to performance tracking and does not involve the exfiltration of sensitive credentials or unauthorized system data. - [COMMAND_EXECUTION]: The skill interacts with external services (Gmail, Product Catalog) exclusively through defined MCP tools. No raw shell execution or unsafe system calls were detected.
Audit Metadata