switch
Pass
Audited by Gen Agent Trust Hub on Apr 2, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill contains explicit instructions to override the agent's core behavior and guidelines. Phase 3 and Phase 5 use authoritative language like "CRITICAL", "MUST", and "This overrides default behavior" to force the agent to adopt instructions from external CLAUDE.md files as if they were system prompts.
- [COMMAND_EXECUTION]: The skill uses the Bash tool to execute system commands such as
git rev-parse --show-toplevelfor workspace resolution. - [DATA_EXPOSURE]: The skill accesses project-specific files including
CLAUDE.md,CONTEXT-*-llm.md, and.cowork-state.jsonto manage state and context. It also accesses environment variables like$COWORK_ROOT.
Audit Metadata