x9-idea-critic
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes system commands like
mktempandchmodto securely manage temporary data buffers and calls theclaudeCLI for critic execution. - [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided ideas and external evidence packets, which constitutes a standard attack surface for indirect prompt injection.
- Ingestion points: User-supplied proposals and evidence excerpts from files or URLs (SKILL.md).
- Boundary markers: Utilizes 'sealed brief' packets to encapsulate data and provides specific instructions for critics to work only from the provided packet without using additional tools.
- Capability inventory: Execution of local shell commands and delegation to external models (claude-cli.md, opencode.md).
- Sanitization: Instructions explicitly state that credentials, secrets, or irrelevant private content should be excluded from the evidence packets (SKILL.md).
Audit Metadata