The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute specific subcommands of the cawplan CLI (e.g., cawplan qa-insights testrail execution summary and cawplan qa-insights testrail execution failures). These commands are used to fetch structured QA data for reporting purposes.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from external TestRail reports, specifically test case names and execution comments, which are interpolated into the agent's output.
Ingestion points: CLI output from execution summaries and failure lists in references/execution.md.
Boundary markers: The instructions mandate the use of markdown tables to structure the report, providing a clear visual delimiter for external content.
Capability inventory: The agent has access to the Bash tool, though its usage is strictly constrained by the instructions to the cawplan command suite.
Sanitization: The skill instructions explicitly require truncating user comments to 80 characters (found in references/execution.md §Failures), which serves as a basic constraint on the volume of injected content, although no content-based filtering is specified.

cawplan-test-execution-progress