mcp-environtment-browser
MCP server acting as a credential vault and browser execution broker, letting AI agents perform browser tasks (login, scraping, etc.) without exposing raw credentials, with per-tab billing and a monitoring UI.
README
mcp-environtment-browser
MCP server untuk credential vault + browser execution broker, dengan subscription per-tab.
Status: Phase 1 COMPLETE (2026-08-16)
Repo ini sudah menyelesaikan Phase 1 (Strategi A: Local-First + License
Server Only) — 9 phase implementation closed, 21 logical commits, 176 unit
tests PASS, 87.24% coverage, hermes verify --json → ok: true.
| Metric | Value |
|---|---|
| Phase completed | 9/9 (Phase 1-9 + bootstrap) |
| Quality gates | 10/10 passed |
| Unit tests | 176 passed, 1 skipped |
| Coverage | 87.24% (gate ≥80%) |
| MCP tools | 13 (per spec §6.4) |
| MCP prompts | 3 (per spec §6.4.1) |
| License server endpoints | 4 (per spec §6.5) |
| Build | wheel 43KB generated |
| Manual validation | 5 items pending owner (init/serve/DB counter/license gate/pause-resume) |
Lihat docs/spec/01_phase1_local_first_license_only/state.json
untuk SSOT fase + quality gates. Lihat docs/spec/01_phase1_local_first_license_only/spec.md
untuk latar belakang, scope, dan strategi implementasi fase 1.
Apa Ini (Goal)
Sebuah MCP server yang memungkinkan agent (Hermes, OpenAI, Claude Code, dll) menjalankan tugas di browser — seperti login ke TikTok, scrape data Google, akses GCP, atau ambil data dari Coretax DJP — tanpa agent harus pegang credential mentah.
Prinsip utama:
- Credential vault — username/password/SSH/OAuth token/env var, disimpan terenkripsi di local user
- Browser execution broker — agent minta "scrape akun TikTok X", MCP yang handle login + ambil data + return JSON
- Per-tab billing — setiap
browser.new_page()dihitung 1 tab, hit server license DB; user tidak terdaftar = tidak bisa pakai - Multi-agent — agent apapun yang bisa connect MCP (stdio transport) bisa pakai, tanpa modifikasi
- 13 MCP Tools + 3 MCP Prompts — Tools untuk runtime action (credential + browser + DevTools + pause/resume + list sessions + web monitoring), Prompts untuk pattern guidance (OAuth confirmation + browser debug + human intervention)
- Web monitoring companion — real-time screenshot tiap tab di
http://localhost:9876, click-to-act untuk user intervene - Cross-platform — dirancang jalan di Linux/Windows/macOS
Stack & Lisensi
| Item | Pilihan |
|---|---|
| Bahasa | Python 3.10+ |
| Server | FastAPI (license DB + per-tab counter) |
| Client | Click/Typer CLI + MCP stdio |
| Browser | Playwright (Python) |
| Credential storage | libsecret/Secret Service API (Linux), extensible ke Windows/macOS |
| Lisensi | MIT |
| Branch default | master |
🚨 Sebelum Kontribusi — WAJIB BACA
AGENTS.md— strict guidelines, anti-over-engineering, owner workflow styleQUICKREF.md— TL;DR orientasi 2 menitdocs/spec/01_phase1_local_first_license_only/spec.md— latar belakang, K1-K7, scope, strategi A
Struktur Docs
.
├── AGENTS.md # ← BACA PERTAMA
├── QUICKREF.md # ← BACA KEDUA (orientasi 2 menit)
├── README.md # file ini
├── LICENSE # MIT
└── docs/spec/01_phase1_local_first_license_only/
├── spec.md # Latar belakang, K1-K7, scope, strategi A
├── state.json # SSOT fase + quality gates
├── knowledge.md # Reuse catatan (cua-driver, libsecret, Playwright)
├── roadmap_migration.md # Strategi A → D → fase 3 (untuk agent masa depan)
└── refactor/
├── 00_overview.md # Quick split matrix + scope contract
├── 10_mcp_server.md # Detail MCP tool spec
├── 20_license_server.md # Detail server endpoint spec
├── 30_client_arch.md # Local-first client structure
├── 40_distribution.md # PyInstaller binary + install script
├── 90_decisions_log.md # CAPTURE K1-K7 — JANGAN DIHAPUS
└── 95_anti_overengineering_guard.md # ← BACA KETIGA (anti-over-engineer)
Status Implementasi
| Phase | Strategi | Status |
|---|---|---|
| Phase 1 | A. Local-first + license-server-only | ✅ COMPLETE (2026-08-16) |
| Phase 2 | D. Hybrid credential-from-server + local browser | 📋 ROADMAP |
| Phase 3 | Multi-tenant + per-target rate limit | 📋 ROADMAP |
Lihat roadmap_migration.md
untuk transisi detail.
Quick Start (setelah install.sh)
# 1. Install (idempotent)
bash install.sh && bash install.sh
# 2. Start license server (di background)
mcp-env-browser-license-server serve --port 8765 &
# 3. Register admin + get API key (sekali)
curl -X POST http://localhost:8765/license/register \
-H "Content-Type: application/json" \
-d '{"email":"owner@mcp-env-browser.local","plan":"dev"}'
# 4. Init client config
mcp-env-browser init --server-url http://localhost:8765 --api-key <key>
# 5. Start MCP stdio + monitoring UI
mcp-env-browser serve
# → MCP stdio untuk agent
# → http://localhost:9876 untuk monitoring UI
Kontribusi
Sampai spec disetujui owner (yadenmustopa), JANGAN tulis kode implementasi.
Tambah/update docs saja.
Setelah spec di-approve, kontribusi mengikuti workflow di
docs/spec/01_phase1_local_first_license_only/refactor/00_overview.md.
Lisensi
MIT — lihat LICENSE.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.