A local bridge that connects coding-agent clients to tools exposed by web pages in Chromium. It uses Puppeteer to control the browser and the Model Context Protocol (MCP) to present those page tools to clients.

Use it to navigate pages, inspect their tool status, run JavaScript in a page, and call tools registered through WebMCP. It is intended for OpenCode, Claude Code, and similar MCP clients.

It removes the need to build a separate connection between an MCP client and each webpage's tools. It also works around a Chromium issue that can prevent page tools from running correctly.

AGENT-BRIEF — Agent by enihcam

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods

npx agentmods add agents/enihcam/opencode-webmcp/agent-brief

Clone the repo

git clone --depth 1 https://github.com/enihcam/opencode-webmcp

Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 492 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. Scan, not verified.

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5	$0.00000	$0.00492
Opus 5	$0.00000	$0.00246
Sonnet 5	$0.00000	$0.00098
Haiku 4.5	$0.00000	$0.00049

Measured yesterday against content hash da488170b8e4, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

AGENT-BRIEF scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

docs/agents/AGENT-BRIEF.md · 46 lines

What it actually says

Agent Brief — opencode-webmcp

Project Overview

Single-file MCP server bridging MCP clients (OpenCode, Claude Code) to Chromium's WebMCP API. Connects via Puppeteer + CDP, discovers tools registered via document.modelContext on the page, and exposes them as MCP tools.

Entry point: server.js (ESM, Node 18+, ~400 lines) Deps: @modelcontextprotocol/sdk + puppeteer-core No build step — edit server.js, restart to apply changes.

Architecture

Bridge-native tools (always available):
- webmcp_navigate — navigate to a URL
- webmcp_status — get page/tool status
- webmcp_evaluate — run JS in page context
- webmcp_invoke_tool — call a registered WebMCP tool
- webmcp_register_test_tools — register test helpers
Page-discovered WebMCP tools — served alongside bridge tools
Monkey-patch via evaluateOnNewDocument: Chrome 150+ has a bug where page.webmcp.invokeTool() returns {}. The bridge intercepts document.modelContext.registerTool to capture each tool's execute function, then replaces executeTool to call captured functions directly.
Auto-recovery: detached-frame errors re-create the page automatically.

Key Implementation Details

Chrome requirement: 150+ with --enable-features=WebMCPTesting,DevToolsWebMCPSupport
puppeteer-core — no bundled browser; must have system Chrome/Chromium
CHROME_PATH, WEBMCP_TARGET_URL, WEBMCP_HEADLESS env vars
--no-sandbox always set; --ozone-platform=wayland added when not headless

Testing

node --check server.js      # syntax check
node test/unit.mjs          # 36 unit tests

Useful Links

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

yesterday First seen · 46 lines · 0 tokens per session scan A da488170b8e4

Subscribe to this mod's changes

AGENT-BRIEF is an agent published in the GitHub repository enihcam/opencode-webmcp (1 stars, last pushed 8d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 492 tokens. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Backend and APIs Data and AI Documents ai-tools browser-automation chrome-devtools chrome-extension claude-code mcp

qa-tester

Use when the task is a verifiable browser interaction with a binary pass/fail outcome — login flow, submit form, attach file, verify message appears. Returns a verdict + evidence. Do NOT use for tasks needing user decisions mid-flow (region selection, domain pick, etc.).

DevZonayed/Mochi · 60 tokens

AGENT-BRIEF

Getting it into your agent

What it costs to keep this loaded

Grade A, and why

Nothing flagged

What it actually says

Agent Brief — opencode-webmcp

Project Overview

Architecture

Key Implementation Details

Testing

Useful Links

What this file has done since we first saw it

Other agents, from other repositories

qa-tester

Autonomous Agents

analyzer

scorer

synthesizer

evaluator-innovation