webmcp-gen — Claude Code skill: WebMCP tools for a site
Writes a WebMCP tool manifest for a specific site, compiles it into an init script, and checks that Chrome registers and invokes the tools.
npx skills add browserbase/skills --skill webmcp-gen -a claude-codeWhat webmcp-gen does
The skill walks you through the loop: pick a slug shaped like <domain>/<task>, scaffold the artifact, explore the page with the browse CLI, describe tools in manifest.json, compile to webmcp.init.js, then validate. There is no nested agent here — you explore the page and edit the manifest yourself, the skill supplies the scripts and the rules.
Under the hood sit scripts/scaffold.mjs, scripts/compile.mjs, scripts/generate-stagehand-example.mjs and scripts/validate.mjs. The generated stagehand-example.ts runs through tsx. Tool code executes inside the page, so implementation.source gets browser APIs only: document, location, navigator and same-origin fetch. Playwright, Puppeteer and Stagehand do not work there — worth checking before you compile.
You need Node 18+, Chrome or Chromium with WebMCP testing flags, and the browse CLI (npm install -g browse). Dependencies install with pnpm install in the skill directory. Keep API keys, tokens, cookies and credentials out of the manifest; eval and new Function are off limits. When validation fails, read eval.json and eval-report.md, patch the manifest, compile and validate again.
What it helps with
- describe WebMCP tools for one site in a manifest
- compile a manifest into
webmcp.init.js - confirm Chrome registers the tools and invokes them
- get a runnable Stagehand example to launch with
tsx - trace a failed validation through
eval-report.md
Example prompts for the agent
- generate a WebMCP tool for example.com that returns the page title and URL
- scaffold the example.com/page-context artifact and compile it to webmcp.init.js
- run validation on the artifact and show me what failed in eval-report.md
How to install webmcp-gen
1Skills CLI (Vercel)
npx skills add browserbase/skills --skill webmcp-gen -a claude-code2Manually: put the files in place
~/.claude/skills/webmcp-gen/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/webmcp-gen/
1Skills CLI (Vercel)
npx skills add browserbase/skills --skill webmcp-gen -a codex2Manually: put the files in place
.agents/skills/webmcp-gen/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add browserbase/skills --skill webmcp-gen -a cursor2Manually: put the files in place
.cursor/skills/webmcp-gen/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install browserbase/skills webmcp-gen2Manually: put the files in place
.github/skills/webmcp-gen/SKILL.md1Skills CLI (Vercel)
npx skills add browserbase/skills --skill webmcp-gen -a gemini-cli2Platform CLI
gemini skills install https://github.com/browserbase/skills --path skills/webmcp-genManually: put the files in place
.devin/skills/webmcp-gen/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add browserbase/skills --skill webmcp-gen -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about webmcp-gen
How do I install webmcp-gen in Claude?
Run npx skills add browserbase/skills --skill webmcp-gen -a claude-code. The skill then lives at ~/.claude/skills/webmcp-gen/SKILL.md.
What does webmcp-gen do?
It helps you describe WebMCP tools for a site in manifest.json, compile them into webmcp.init.js, and verify that Chrome registers and invokes the tools.
What does webmcp-gen need to run?
Node 18+, Chrome or Chromium with WebMCP testing flags, and the browse CLI installed via npm install -g browse. Skill dependencies come from pnpm install in its directory.
Source text
SKILL.md · MIT
---
name: webmcp-gen
description: Create, compile, and validate site-specific WebMCP init scripts from a target URL and desired tool capability. Use when the user wants to author WebMCP tools for a website, produce a webmcp.init.js artifact, or test WebMCP registration and invocation through Stagehand.
compatibility: "Requires Node 18+ and Chrome/Chromium with WebMCP testing flags. Run `pnpm install` in the skill directory to install the Stagehand dependency. Page exploration uses the browse CLI (`npm install -g browse`)."
license: MIT
allowed-tools: Bash Read Grep Edit Write
---
# WebMCP Gen
Author website-specific WebMCP tools by writing a manifest, compiling it to an init script, and validating that Chrome registers and invokes the tools.
This skill does not call a nested agent. You are responsible for exploring the page, writing `manifest.json`, and iterating based on validation output.
## Setup check
From the skill directory, install dependencies if they are not already installed:
```bash
cd skills/webmcp-gen
pnpm install
```
This installs the pinned Stagehand package plus the TypeScript toolchain (`tsx`,
`typescript`, `@types/node`) used to run the generated `stagehand-example.ts`.
## Workflow
1. Pick an artifact slug with exactly one slash:
```text
<domain>/<task>
```
Example:
```text
example.com/page-context
```
2. Scaffold the artifact:
```bash
node scripts/scaffold.mjs example.com/page-context --url https://example.com
```
3. Explore the target page with the `browse` CLI:
```bash
browse open https://example.com --local
browse snapshot
browse get title
browse get url
browse get text body
browse get html body
```
Prefer `browse snapshot`, page text, and DOM inspection over screenshots unless visual layout matters. Use `browse stop` when exploration is complete.
4. Edit `artifacts/<domain>/<task>/manifest.json`. The manifest is the source of truth.
5. Compile:
```bash
node scripts/compile.mjs artifacts/example.com/page-context
```
6. Generate a runnable Stagehand example (`stagehand-example.ts`) and run it with `tsx`:
```bash
node scripts/generate-stagehand-example.mjs artifacts/example.com/page-context
npx tsx artifacts/example.com/page-context/stagehand-example.ts
```
7. Validate:
```bash
node scripts/validate.mjs artifacts/example.com/page-context
```
8. If validation fails, inspect `eval.json` and `eval-report.md`, patch `manifest.json`, then compile and validate again.
## Manifest contract
```json
{
"domain": "example.com",
"task": "page-context",
"url": "https://example.com",
"generatedAt": "2026-06-04T00:00:00.000Z",
"tools": [
{
"name": "example_com_page_context",
"description": "Returns page context.",
"inputSchema": {
"type": "object",
"properties": {},
"required": []
},
"implementation": {
"kind": "dom",
"source": "return { success: true, title: document.title, url: location.href };"
},
"fixtureInput": {}
}
]
}
```
## Authoring rules
- `implementation.source` is inserted inside `async (input) => { ... }`; write JavaScript statements, not a full function wrapper.
- Return a JSON-serializable object.
- WebMCP code runs inside the browser page. Use browser-native APIs: `document`, `location`, `navigator`, and same-origin `fetch`.
- Do not use Playwright, Puppeteer, Stagehand, XPath helpers, or agent/browser commands inside `implementation.source`.
- `document.querySelector` and `querySelectorAll` must receive valid browser CSS selectors only.
- To find visible text, use `Array.from(document.querySelectorAll(...)).find((el) => (el.textContent || "").includes("..."))`.
- Do not include API keys, bearer tokens, cookies, localStorage secrets, or user credentials.
- Do not use `eval` or `new Function`.
- Avoid destructive actions unless the user explicitly asked for them.
- Make implementations defensive: check for missing elements and return structured `{ success: false, error: "..." }` responses.
- Generated init scripts register WebMCP tools only in the top frame.
## Output layout
```text
artifacts/<domain>/<task>/
manifest.json
webmcp.init.js
stagehand-example.ts
eval.json
eval-report.md
```
To turn the example into a standalone project, scaffold a Stagehand app with
`npx create-browser-app` and drop the generated `webmcp.init.js` into it (load it
with `page.addInitScript({ path: "webmcp.init.js" })`).
© browserbase · MIT · browserbase/skills