The Agent Skills Directory

[SAFE]: No security issues detected. The skill follows best practice guidelines for tool registration, emphasizing user interaction for sensitive operations and schema validation for inputs.
[EXTERNAL_DOWNLOADS]: Fetches technical specifications and documentation from the Web Machine Learning GitHub Pages repository, which is an established source for this technology.
[INDIRECT_PROMPT_INJECTION]: The skill describes an interface for processing agent-supplied data.
Ingestion points: execute(input, client) callback in navigator.modelContext.registerTool (SKILL.md).
Boundary markers: The API architecture uses inputSchema for JSON Schema validation of all inputs.
Capability inventory: Code examples demonstrate standard browser fetch usage for data retrieval and submission.
Sanitization: Recommends using inputSchema for validation and client.requestUserInteraction for human-in-the-loop authorization of destructive actions.

webmcp-register-tool