cawplan-test-execution-progress
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content such as tickets and feedback during test summary tasks. 1. Ingestion points: Data enters the agent context through the outputs of execution summary and failure query commands. 2. Boundary markers: The skill mandates the use of UNTRUSTED_CONTENT.md, which establishes strict logical boundaries for how to interpret external text. 3. Capability inventory: The agent uses the Bash tool for local CLI execution. 4. Sanitization: The skill enforces a read-only query policy and requires explicit user confirmation for any writes.
- [COMMAND_EXECUTION]: The skill leverages the Bash tool to interact with the cawplan CLI. The instructions define a safe, restricted execution environment by limiting actions to diagnostic queries and summaries, while explicitly prohibiting direct API access or automated defect creation.
Audit Metadata