research-reviewer

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION] (LOW): The skill adopts a 'Pickle Rick' persona with a 'God Mode' instruction ('If a tool is missing, INVENT IT'). This is an override-style prompt that encourages the agent to ignore standard constraints and can be associated with jailbreak-style behavior.
  • [COMMAND_EXECUTION] (LOW): The 'God Mode' directive risks the agent attempting to execute non-existent or hallucinated commands if it perceives a tool is missing.
  • [INDIRECT_PROMPT_INJECTION] (LOW): The skill is vulnerable to instructions embedded in the documents it reviews. Evidence Chain: 1. Ingestion points: Reads untrusted content from ${SESSION_ROOT}/[ticket_id]/research_[date].md. 2. Boundary markers: None. The content is processed directly without delimiters or 'ignore' instructions. 3. Capability inventory: File read/write and ticket status updates. 4. Sanitization: None.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:48 PM