SciREPL MCP
MCP servers for controlling SciREPL from coding agents and MCP clients, offering an app-connected broker and a Playwright driver for browser automation.
README
SciREPL MCP
SciREPL MCP contains two independently installable servers for controlling SciREPL from coding agents and other Model Context Protocol clients. Choose the server that matches where SciREPL is running; installing one does not install the other's dependencies.
| Server | Best for | Connection | Browser dependency |
|---|---|---|---|
| App-connected broker | SciREPL Pro on Android or in a browser | SciREPL connects outward to the broker; MCP clients use Streamable HTTP | None |
| Playwright driver | Free-PWA, hosted-PWA, and local-browser workbook automation | The server launches or attaches to Chromium over CDP; Android attachment is planned | Playwright and Chromium |
The two approaches are complementary. The broker calls the app's own approved tools and can optionally host a remote coding-agent or terminal session. The Playwright driver operates the visible notebook interface and is useful for automated workbook testing, screenshots, and diagnosing browser behaviour.
App-connected broker
git clone https://github.com/s243a/SciREPL-MCP.git
cd SciREPL-MCP
./setup-broker.sh
~/scirepl-broker/start-broker.sh
Setup requires Node.js 20 or newer and fails before writing if the current shell
still points to an older system Node. If you use NVM, run nvm use 22 first. The
generated launcher retains that validated Node executable.
On native Windows, the core bridge can be prepared with
./setup-broker.ps1 from PowerShell. Agent and terminal modes currently use
POSIX process conventions and should run under WSL. Native PowerShell/ConPTY is
a possible future terminal profile, not a current feature.
The broker is executed directly by Node.js; it is not compiled. Setup installs
only its JavaScript dependencies by default. Optional terminal mode installs and
verifies the native node-pty dependency on the machine that will run it.
See the broker guide for pairing SciREPL Pro, connecting an MCP client, and enabling the optional agent or terminal surfaces.
Playwright driver
git clone https://github.com/s243a/SciREPL-MCP.git
cd SciREPL-MCP/packages/playwright-driver
npm ci
npx playwright install chromium
npm test
npm start
See the Playwright driver guide for MCP client configuration, attaching to an existing Chromium session, and the tool catalog.
Repository checks
The root package is an orchestration convenience only. It is deliberately not an npm workspace, so installing the broker never pulls in Playwright.
npm run install:all
npm test
npm run audit
npm run setup is the explicit broker-environment setup command, not the
repository dependency-install command.
Security and privacy
The app-connected broker binds to loopback by default and requires a pairing token. Its remote-agent and terminal endpoints are disabled by default. The Playwright driver can control a browser profile and everything visible to that profile. Treat both as trusted local development tools, use encrypted tunnels for remote access, and read the security policy before exposing either service.
Neither server adds analytics or telemetry. Hosted model clients and coding agents may still send selected notebook content to their configured providers.
The repository stores agent guidance only as inert .template files. Explicit
agent setup materializes conventional AGENTS.md, CLAUDE.md, and provider
settings inside a dedicated private session workspace; ordinary broker startup
does not create them. See the agent-context guide.
Project relationship
This standalone repository is intended to be the canonical public home for both
SciREPL MCP servers. After its initial publication it can be referenced from
UnifyWeaver at examples/sci-repl/mcp as a Git submodule, without mixing its
history into the UnifyWeaver repository.
The browser driver was originally developed in the public UnifyWeaver project; the app-connected broker was extracted from SciREPL Pro. See PROVENANCE.md for the exact source revisions and boundaries.
License
MIT. See LICENSE. Direct dependency licensing is summarized in THIRD_PARTY_NOTICES.md.
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.