The Agent Skills Directory

[COMMAND_EXECUTION]: The skill generates and executes shell scripts in the tests/qa/ directory and utilizes system tools like curl, grpcurl, and httpie as described in references/README.md.\n- [REMOTE_CODE_EXECUTION]: The persona performs dynamic code execution by writing bash scripts to the filesystem and executing them after applying chmod +x.\n- [EXTERNAL_DOWNLOADS]: The use of curl and httpie enables the agent to interact with network services, which could include external remote resources.\n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8). 1. Ingestion points: API responses from curl and grpcurl commands in references/README.md. 2. Boundary markers: None present. 3. Capability inventory: Execution of generated bash scripts and Git operations. 4. Sanitization: No evidence of response validation or sanitization before processing.

qa-engineer