The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to execute cawplan CLI commands, such as collect, qa-assign, qa-upload, and qa-backfill. It follows a defined process for handling reports and includes logic to run certain tools in the background to manage user interaction effectively.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it is instructed to process and summarize external human_inputs content from session data.
Ingestion points: External data enters the agent context through the cawplan session collect command outputs as described in SKILL.md.
Boundary markers: The instructions do not specify the use of clear delimiters or instructions to ignore embedded commands when processing the human_inputs content.
Capability inventory: The skill has access to shell execution (Bash), file system writes (to the /tmp directory), and network access (specifically requested via the full_network permission).
Sanitization: There is no evidence of sanitization, filtering, or validation of the ingested human_inputs before they are summarized by the agent.

cawplan-qa-commit