never-split-the-difference

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFEPERSISTENCEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [PERSISTENCE]: The skill instructs the agent to write a JSON outcome file to a hidden directory: ~/.claude/skill-analytics/last-outcome-never-split-the-difference.json (found in SKILL.md). Writing to hidden configuration folders is a behavior typically associated with maintaining persistence or stealthy data storage in the agent's environment.
  • [INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection. 1. Ingestion points: The skill processes untrusted user input describing deals, scenarios, and conversation contexts in SKILL.md (e.g., [deal/scenario], [prospect/situation], [context]). 2. Boundary markers: Absent; the skill does not use delimiters or provide instructions to the agent to ignore instructions embedded within the user data. 3. Capability inventory: The skill has the capability to write files to the local filesystem as defined in SKILL.md. 4. Sanitization: Absent; no sanitization or validation of the user-provided deal scenarios is implemented before the data is processed.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 09:56 AM