Let AI agents work with test suite your team owns
Use your AI agent to plan, create, run and troubleshoot tests. BugBug handles execution while your team controls test intent.
WHAT THE AGENT CAN DO
Tools to work across the full regression cycle
BugBug MCP gives your agent access to the projects, tests, suites, variables, runs and failure evidence already managed in BugBug.
Tests
List, read, create, edit, import, reorder and run tests.
Suites
Build and update suites, then trigger them on demand.
Runs
Start runs, watch progress live, stop them and read results.
Steps
Create, update and delete individual steps with typed payloads.
Groups & components
Convert groups to reusable components and manage their usage.
Debugging
Inspect failed step details, screenshots, logs and query the recorded DOM snapshot.
Visual regression
Review reference screenshots and resolve visual differences.
Profiles & variables
Read profiles and manage environment variables safely.
Project & docs
Read project settings, upload artifacts and ask the BugBug docs.
USE CASES
One regression suite. More ways to get work done.
Start with a prompt. The agent uses BugBug MCP to do the work in your existing project.
Plan test coverage
Review your application context and existing BugBug tests, then identify and prioritize potential coverage gaps.
Use suggested promptCreate a test from a description
Describe a user flow in plain language and get a runnable BugBug test with steps and assertions.
Use suggested promptDebug a failed run
The agent pulls failed step details, logs and the DOM snapshot, then explains the root cause.
Use suggested promptFix a broken selector
Repair unstable locators using the recorded DOM instead of guessing in the UI.
Use suggested promptRefactor existing tests
Turn repeated step groups into reusable components and clean up naming across tests.
Use suggested promptCreate a comprehensive report
Get a read-only summary of flaky tests, recent failures and coverage for your standup.
Use suggested promptBUGBUG SKILLS
MCP gives access. Skills guide how the agent tests.
Skills are BugBug-specific instructions that teach your agent how to test properly: when to inspect, when to ask, when to run, and when to stop.
npx skills add https://github.com/bugbug-io/agent-pluginContext discovery
Learns your project, repo and tested app before doing anything else.
Tests planning
Plans E2E and UI coverage before writing exact steps.
Tests authoring
Creates, edits and validates complete tests.
Selectors authoring
Writes, reviews and repairs stable DOM selectors.
Test run debugging
Diagnoses failures from real run evidence.
YAML authoring
Creates, validates and repairs BugBug YAML before import.
Tests refactoring
Reviews and refactors tests, groups, components and suites.
Project reporting
Builds read-only project health reports for QA and PMs.
HOW TO CONNECT
Connected in under a minute
Pick your AI client, run one command and sign in.
Install the plugin
One command installs BugBug MCP and BugBug Skills in your AI client.
Sign in with OAuth
Restart the client and complete the BugBug sign-in flow.
Verify the connection
Ask "List my BugBug projects" to confirm the connection is working.
npx @bugbug-io/cli plugin --agent=claudeFAQ
Questions you may have
The BugBug MCP server connects supported AI agents to your BugBug projects. It lets them plan coverage, create and maintain tests, run suites, and investigate failures using real test data and run evidence, all through natural-language requests.