Debugg AI MCP Server

Community

Zero-Config, Fully AI-Managed End-to-End Testing

How to install ↓ Compare GitHub source repository ↗

Category: Other #88 of 230
Stars: ★ 68 Popular
Transport: stdio (local process) · Streamable HTTP
Runtime: Node.js 20.20.0+ · Docker
Credentials: API key / credential required
License: Apache-2.0
Last commit: 6 days ago
Tools: 8

58FMRS · C

The Debugg AI MCP server provides a comprehensive AI-driven browser testing suite, suitable for teams needing automated end-to-end tests with integrated test management.

Strongest · Documentation 16/20 Weakest · Reliability 6/20

Reliability

6/20

Security and permissions

10/20

Maintenance

14/20

Documentation

16/20

Setup experience

12/20

Why each score

Reliability 6/20: No verifiable execution evidence. The repository may have tests, but from the README and static files alone, it's unclear if the server starts cleanly. Tool definitions are detailed, but actual behavior alignment can't be confirmed. Deducted for lack of reproducible execution evidence and error-handling tests.
Security and permissions 10/20: No malware or credential leaks evident. Passwords are write-only, tunnel URLs are stripped, deletes require confirmation, and API key is passed via environment variable. However, not all code paths were audited, and external network dependencies (ngrok, caddy) and remote transport authorization details need scrutiny.
Maintenance 14/20: Repository is active, not archived, has Apache-2.0 license. Few open issues, but no evidence of security response channel or release cadence. Deducted for lack of commit history and dependency update records.
Documentation 16/20: Documentation is very comprehensive, covering installation, auth, tool parameters, limitations, migration, and troubleshooting, with security invariants. But some content goes beyond verifiable evidence, like multi-replica deployment details, which may be implementation assumptions.
Setup experience 12/20: Provides npm and Docker install examples, but depends on Node.js version and Caddy, and Dockerfile has known gaps. Local dev steps are clear, but no CI or test evidence, so setup may be fragile.

Static review · not runListed 2026-08-07

Read the FMRS scoring method →

Fit and risk

What it can accessUses the networkChanges third-party account dataControls a browser

Best for

Development teams wanting to automate end-to-end testing of web applications
CI pipelines that need quick page sanity checks
Users of the DebuggAI platform who want MCP integration

Not for

Unit tests or integration tests that don't require AI
Offline environments that cannot access the service (requires API key)
Fully local or self-hosted tests that don't rely on external services

Required permissions

Requires internet access to the DebuggAI backend and an API key
Requires the ability to allow an AI agent to perform browser interactions
May require tunneling for localhost access (via ngrok/caddy)

Risks and side effects

AI agent behavior may affect production if run against production
Testing depends on external service and API key, potential availability issues
Credential handling: passwords are write-only, but environment variables must be stored securely

Setup

Before you start

Runtime：Node.js 20.20.0+ · Docker

DEBUGGAI_API_KEY requiredsecret Required DebuggAI backend API key, get one at debugg.ai; aliases DEBUGGAI_API_TOKEN and DEBUGGAI_JWT_TOKEN are accepted.

POSTHOG_API_KEY optionalsecret Optional override of the embedded telemetry project key (e.g. for a private fork).

Other optional settings (10)

DEBUGGAI_API_URL optional Optional backend API base URL, defaults to https://api.debugg.ai.

DEBUGGAI_TOKEN_TYPE optional Optional, token (default) or bearer; set to bearer for remote HTTP transport.

DEBUGGAI_EVAL_TEMPLATE optional Optional override of the App Evaluation workflow slug dispatched by check_app_in_browser, default flow/e2es/app-eval.

LOG_LEVEL optional Optional log level: error / warn / info (default) / debug.

DEBUGGAI_TELEMETRY_DISABLED optional Optional; set to 1/true/yes/on to disable telemetry entirely.

CADDY_BIN optional Optional path to a local Caddy binary; only needed if automatic download failed and you test localhost URLs.

DEBUGGAI_MCP_TRANSPORT optional Optional, stdio (default) or http for the hosted remote transport.

PORT optional Optional HTTP listen port for the http transport, default 3000.

DEBUGGAI_MCP_PUBLIC_URL optional Optional public resource URL for this server (RFC 9728), default https://mcp.debugg.ai.

DEBUGGAI_OAUTH_ISSUER optional Optional authorization server advertised to clients, default https://auth.debugg.ai.

Get an API key at debugg.ai.
Add the server to your MCP client config using npx and set the DEBUGGAI_API_KEY environment variable.
Alternatively, use Docker with the provided image (note: the current image may lack support for localhost tools).
For remote HTTP transport, set DEBUGGAI_MCP_TRANSPORT=http and other optional variables.

claude_desktop_config.json

{
  "mcpServers": {
    "debugg-ai": {
      "command": "npx",
      "args": ["-y", "@debugg-ai/debugg-ai-mcp"],
      "env": {
        "DEBUGGAI_API_KEY": "your_api_key_here"
      }
    }
  }
}

Shown for Claude Desktop. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.

.vscode/mcp.json

{
  "servers": {
    "debugg-ai": {
      "command": "npx",
      "args": [
        "-y",
        "@debugg-ai/debugg-ai-mcp"
      ],
      "env": {
        "DEBUGGAI_API_KEY": "your_api_key_here"
      }
    }
  }
}

Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).

Terminal

claude mcp add debugg-ai -e DEBUGGAI_API_KEY=your_api_key_here -- npx -y @debugg-ai/debugg-ai-mcp

Run it in a terminal; replace any <…> placeholders with your own values first.

Generate and validate client config

Check that it works

The client tool list should show all 8 tools (check_app_in_browser, trigger_crawl, probe_page, project, environment, test_suite, test_case, executions); calling project {action:"list"} and getting paginated results proves the API key and connection work.

Troubleshooting

Ensure DEBUGGAI_API_KEY is set and valid
Check Node.js version (20.20.0+)
For localhost URLs, ensure the caddy binary is available (or set via CADDY_BIN)
Verify that the local service is running if necessary

Things to try

Once connected, you can ask your AI assistant things like:

Run the AI browser agent against http://localhost:3000 to test the login flow with my test account and return screenshots
Do a lightweight no-LLM probe of /settings and /dashboard to check whether my refactor broke anything
Trigger a crawl of localhost:3000 to populate the project knowledge graph
List my test suites and run the one with the lowest pass rate

Tools 8

check_app_in_browser writes

Runs an AI browser agent to test your app based on a natural language description, with automatic tunneling for localhost URLs.

trigger_crawl writes

Triggers a server-side browser-agent crawl to populate the project's knowledge graph.

probe_page read-only

Lightweight no-LLM batch page probe that checks 1-20 URLs for rendered state, console errors, and network summary.

project read-only

Manage projects: get, list, create.

environment writes

Manage environments and credentials, including login session operations.

test_suite writes

Manage test suites: list, create, run, get results, delete.

test_case writes

Manage test cases: create, update, delete.

executions read-only

Retrieve execution details and artifacts.

Use cases

End-to-end AI-driven testing of web apps based on natural language descriptions

Lightweight page-based smoke tests after each pull request without LLM overhead

Managing test suites, cases, and environments for continuous testing

Debugging runtime issues with HAR and console log capture

Supported clients

Claude Desktop

Claude Code

Listed from the project's documentation, not tested by this site.

Overview

The Debugg AI MCP server provides AI-powered browser testing via the Model Context Protocol. Point it at any URL (or localhost) and describe what to test — an AI agent will browse your app and return pass/fail with screenshots. The server offers 8 tools across browser testing, project management, environment setup, test suite and case management, and execution tracking. It supports multi-entity operations (projects, environments, test suites, test cases, executions) and includes a lightweight page probe for no-LLM smoke tests. Requires Node.js 20.20.0 or later and an API key from debugg.ai.

Similar servers

Hevy MCP 77 · B

Manage your Hevy workout data from AI assistants via MCP.

★ 509 · Tools 26 Compare with this →

Firecrawl MCP Server 75 · B

Firecrawl's official MCP server for web search, scraping, and structured extraction for AI agents

★ 7.6k · Tools 11 Compare with this →

BioMCP 75 · B

One binary. One grammar. Evidence from the biomedical sources you already trust.

★ 654 · Tools 11 Compare with this →

Compartment 75 · B

Encrypted, fully offline long-term memory for AI agents

★ 579 · Tools 14 Compare with this →

Compare with other MCP servers

Source revision 9c626c72e87d Data synced 2026-10-12 Read the FMRS scoring method