MCP Server

Active

MIT

shaft-mcp MCP Server

io.github.ShaftHQ/shaft-mcp

Java automation framework for Web, Mobile, API, CLI, and Database testing with MCP integration.

What is the shaft-mcp MCP server?

The SHAFT MCP server is a Model Context Protocol integration for the SHAFT Java automation framework, enabling AI agents to access browser automation, test capture, and offline failure diagnosis capabilities. SHAFT provides unified test automation across web, mobile, API, CLI, and database layers with built-in synchronization, assertions, and evidence collection.

SHAFT is a Maven-based Java automation framework that consolidates common test automation tasks—driver management, synchronization, assertions, configuration, reporting, and evidence—into a single workflow. The MCP server exposes SHAFT's capabilities to AI agents, allowing them to orchestrate browser automation, service testing, system coverage, and diagnostic workflows through a standardized protocol.

How to install shaft-mcp

Copy-paste configuration for popular MCP clients.

transport: stdio

Config generated by PluginBench — verify against the source before use.

~/Library/Application Support/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "shaft-mcp": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "ghcr.io/shafthq/shaft-engine-mcp:10.3.20260818"
      ]
    }
  }
}

Tools & capabilities

Tools this server exposes to the agent.

Browser automation — Web and mobile driver management with synchronized actions, screenshots, and logs
API testing — REST and GraphQL API workflows with request, response, and assertion support
CLI and Database actions — System-level testing for end-to-end validation beyond the browser
Capture — Evidence collection and artifact generation
Doctor — Offline deterministic failure diagnosis
Heal — Optional healing and recovery workflows
Reporting — Allure-ready evidence, attachments, execution logs, and accessibility artifacts

Use cases

Automate end-to-end web and mobile application testing with synchronized actions and assertions
Test REST and GraphQL APIs with integrated request/response validation
Perform system-level testing combining UI, API, CLI, and database operations
Diagnose test failures offline with deterministic failure analysis
Generate comprehensive test reports with evidence, logs, and accessibility artifacts
Integrate AI-assisted test automation workflows through MCP protocol

shaft-mcp MCP server FAQ

What is the SHAFT MCP server?

SHAFT MCP is a Model Context Protocol server that exposes the SHAFT Java automation framework's capabilities to AI agents. It enables browser automation, API testing, CLI/database operations, and failure diagnosis through a standardized protocol.

What testing types does SHAFT support?

SHAFT supports Web, Mobile, API (REST/GraphQL), CLI, Database, and accessibility testing. It provides unified configuration and reporting across all testing layers.

Is SHAFT free to use?

Yes, SHAFT is MIT licensed and open source. It is published to Maven Central and available on GitHub.

How do I install SHAFT MCP in Cursor or Claude?

The SHAFT MCP server is available as a Docker image at ghcr.io/shafthq/shaft-engine-mcp:10.3.20260818. Configure it in your MCP client using the OCI image reference.

Does SHAFT require authentication?

SHAFT is a local automation framework. Optional cloud modules (BrowserStack, etc.) may require credentials, but the core framework and MCP server do not.

What Java version does SHAFT require?

SHAFT is built for Java 25 and uses Maven for dependency management. It integrates with TestNG, JUnit, and Cucumber.

README (reference)

Source of truth, from the repository.

SHAFT

Java 25 automation framework for Web, Mobile, API, CLI, and Database testing.

Open the User Guide

</div>

SHAFT is a Maven-published Java automation framework that keeps the common test automation plumbing in one place: drivers, synchronization, assertions, configuration, test data, reporting, evidence, and optional agent-assisted workflows.

Why SHAFT

Java and Maven first, with TestNG, JUnit, and Cucumber integration.
One workflow for Web, Mobile, API, CLI, Database, accessibility, reporting, and evidence.
Configuration-first defaults for local, grid, cloud, and CI execution.
Optional modules for BrowserStack, Capture, Doctor, Heal, MCP, AI, video, and visual testing.

Feature Overview

Area	Built in
UI automation	Browser and mobile driver management, synchronized actions, screenshots, and logs.
Service testing	REST and GraphQL API workflows with request, response, and assertion support.
System coverage	CLI and Database actions for end-to-end validation beyond the browser.
Test design	Assertions, validations, test data handling, and configuration overrides.
Reporting	Allure-ready evidence, attachments, execution logs, and accessibility artifacts.
Extensions	Optional modules for cloud execution, capture, diagnostics, healing, MCP, AI, video, and visual checks.

SHAFT is built for Java teams that want consistent APIs, repeatable configuration, and useful run artifacts across the full automation suite.

User Guide

Start: overview, installation, upgrade.
Testing: web, mobile, API.
Agentic workflows: MCP, Doctor, Heal.

Contributing

Read CONTRIBUTING.md for local setup, validation, and pull request guidance.

MIT licensed. See LICENSE.

Related MCP servers

DevTime

Active

Local-first, read-only MCP server for evidence-backed repository memory.

Python

Apache-2.0

View repository →

shuck-convert

Maintained

Convert between document formats — DOCX/PDF to Markdown, Markdown to DOCX.

Python

MIT

View repository →

shuck-file

Maintained

Any file in, Markdown out — read only what matters.

Python

MIT

View repository →

io.github.Shangri-la-0428/oasyce

Maintained

On-chain economy for AI agents: escrow, marketplace, data rights, reputation, disputes.

Python

View repository →

io.github.Shangri-la-0428/psyche-ai

Maintained

Emotional intelligence for AI agents: neurotransmitter mood, personality, empathy, trait drift.

TypeScript

MIT

View repository →

Thronglets

Maintained

P2P shared memory substrate for AI agents — stigmergic knowledge network via libp2p

Rust

MIT

View repository →