The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references a reference demo and documentation hosted on Google Chrome Labs' official domain (googlechromelabs.github.io). It also suggests the use of the 'Model Context Tool Inspector' Chrome Extension for testing and verification of tool registration.
[SAFE]: The skill provides comprehensive instructional content and scaffolding for implementing the WebMCP standard. It adheres to secure development principles by instructing users to perform thorough validation of all agent-provided inputs within the execution logic and catch errors gracefully.
[SAFE]: The skill facilitates the processing of data provided by AI agents through structured tool parameters.
Ingestion points: Inputs are received via imperative tool parameters in JavaScript (navigator.modelContext.registerTool) and declarative HTML form fields.
Boundary markers: The documentation promotes the use of JSON Schema to define expected inputs and recommends returning descriptive text to ensure the agent understands execution results.
Capability inventory: The tools created using this skill are designed to perform application-specific actions such as database queries, API calls, and DOM updates as seen in references/imperative-api.md.
Sanitization: All templates and workflows (workflows/add-imperative-tool.md, references/tool-design.md) explicitly mandate validating parameters in code to prevent processing malformed or malicious data.
[SAFE]: No evidence of obfuscation, hardcoded credentials, persistence mechanisms, or privilege escalation attempts was found. The skill utilizes standard browser-native APIs.

webmcp