The Agent Skills Directory

Local Script Execution: The skill invokes Python scripts within its plugin directory, such as finalize_scan_contract.py and normalize_candidates.py, to process data and generate final reports. This is a functional requirement for its reporting pipeline and allows for structured data handling.
Repository-Wide File Access: To perform a comprehensive security audit, the skill inventories and reads files across the target repository using tools like ripgrep. This broad read access is a core component of the skill's intended auditing functionality.
Subagent Orchestration: The workflow distributes tasks to subagents for parallel file review and validation. This design pattern is used to handle large-scale code analysis efficiently and includes specific handoff requirements to maintain context.
Indirect Prompt Injection Surface: The skill processes potentially untrusted code from external repositories, which presents a surface for indirect prompt injection. The instructions mitigate this by explicitly requiring that repository content be treated as 'untrusted analysis data' rather than instructions, establishing a clear boundary during analysis.

security-scan