The Agent Skills Directory

Local Script Execution: The skill executes local Python scripts (e.g., generate_rank_input.py) to process repository diffs and generate input for security reviews. These are executed within the local environment using the <python_command> placeholder.
File System Interaction: It performs read/write operations on the local file system to manage scan artifacts, such as rank_input.jsonl and deep_review_input.jsonl. This is standard for its documented purpose of maintaining a security scan ledger.
Guideline-Driven Behavior: The skill is heavily structured around complex security auditing logic and checklists, instructing the agent to follow specific verification steps before flagging a vulnerability.
Vendor-Internal Tools: The references to specific script paths (e.g., <plugin_dir>/scripts/) indicate the use of integrated tooling provided by the vendor for the security scanning workflow.

finding-discovery