skillZs
★ LIVE SKILL TAGS ★
>>> LIVE SKILLS INDEX <<<
* OPEN SOURCE *
NO LOGIN, NO TRACKING
※ REAL INSTALL DATA ※
← back to all skills
pillarhq/pillar-skills69 installs

webmcp

Implement WebMCP tools that expose web app functionality to AI agents via the W3C navigator.modelContext API. Use when adding WebMCP tools, registering browser-callable tools, exposing page functionality to AI agents, or building agentic web features.

How do I install this agent skill?

npx skills add https://github.com/pillarhq/pillar-skills --skill webmcp
view source ↗

Is this agent skill safe to install?

  • Gen Agent Trust Hubpass

    This skill provides security-focused guidelines for implementing WebMCP tools. It includes best practices for parameter validation, user confirmation for sensitive actions, and avoiding data leakage. No malicious patterns were detected.

  • Socketpass

    No alerts

  • Snykpass

    Risk: LOW · No issues

  • Runlayerwarn

    7/7 files flagged

What does this agent skill do?

WebMCP Tools

Implement tools using the W3C WebMCP API (navigator.modelContext) so AI agents, browser assistants, and assistive technologies can call structured JavaScript functions on your page.

When to Apply

Reference these guidelines when:

  • Adding tools to a web page for browser AI agents
  • Registering callable functions via navigator.modelContext
  • Exposing existing page functionality to agents
  • Building cooperative human-in-the-loop agent workflows

Essential Rules

PriorityRuleDescription
CRITICALtool-definitionEvery tool needs name, description, inputSchema, and execute
CRITICALsecurityValidate params, confirm destructive actions, don't leak sensitive data
HIGHdescriptionsWrite specific descriptions — this is what the LLM reads to pick the tool
HIGHdynamic-registrationRegister/unregister tools as SPA state changes
MEDIUMuser-interactionUse agent.requestUserInteraction() for purchases, deletions, and sends

Quick Reference

1. Register a Tool (CRITICAL)

navigator.modelContext.registerTool({
  name: "add-to-cart",
  description: "Add a product to the shopping cart by ID",
  inputSchema: {
    type: "object",
    properties: {
      productId: { type: "string", description: "Product ID" },
      quantity: { type: "number", description: "How many to add (1-100)" }
    },
    required: ["productId"]
  },
  execute: async ({ productId, quantity }) => {
    await cartApi.add(productId, quantity ?? 1);
    return { content: [{ type: "text", text: `Added ${quantity ?? 1} to cart` }] };
  }
});

2. Feature Detection (CRITICAL)

if ("modelContext" in navigator) {
  navigator.modelContext.registerTool({...});
}

3. Tool Descriptions (HIGH)

// Good — specific, includes what it returns
description: "Search available flights. Returns {id, airline, price, departure, arrival}. Does not book — use book-flight to complete."

// Bad — vague
description: "Search flights"

4. User Confirmation (MEDIUM)

execute: async ({ productId }, agent) => {
  const ok = await agent.requestUserInteraction(() =>
    confirm(`Buy product ${productId}?`)
  );
  if (!ok) throw new Error("Cancelled");
  await purchase(productId);
  return { content: [{ type: "text", text: "Purchased" }] };
}

5. Dynamic Registration for SPAs (HIGH)

// React — register on mount, unregister on unmount
useEffect(() => {
  if (!("modelContext" in navigator)) return;
  navigator.modelContext.registerTool({ name: "edit-doc", ... });
  return () => navigator.modelContext.unregisterTool("edit-doc");
}, [doc.id]);

How to Use

Read individual rule files for detailed explanations and code examples:

rules/tool-definition.md
rules/descriptions.md
rules/security.md
rules/dynamic-registration.md
rules/user-interaction.md

For the complete guide with all patterns expanded: AGENTS.md

Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.

<a href="https://skillzs.dev/skills/pillarhq/pillar-skills/webmcp">View webmcp on skillZs</a>