The Agent Skills Directory

[SAFE]: No security issues detected. The skill instructions focus entirely on accessibility auditing tasks using official vendor tools (mcp__accesslint__*).
[SAFE]: The permission to use tools like Read, Edit, Write, and Bash is within the expected scope for a skill designed to 'audit and fix' source code. The instructions guide the agent to perform these actions only upon user request and for the specific purpose of resolving accessibility violations.
[SAFE]: Indirect prompt injection surfaces exist as the tool processes external HTML and live DOM content. However, this is an inherent risk for analysis tools. The instructions provide guidance on using structured rule IDs and verifiable 'Source:' lines which helps maintain focus on valid data during the audit process.
Ingestion points: External URLs analyzed via audit_live and local files read via Read tool.
Boundary markers: None explicitly defined in the instructions.
Capability inventory: Edit, Write, and Bash tools allow for file modification and command execution.
Sanitization: No specific sanitization methods are described for the external content before it is processed.

audit