Bugbug
Bugbug is a low-code test automation tool. The easiest way to cut costs of software testing. Automate your web app testing 3x faster without scaling your QA team.
Verdict
Common use cases
- Verify deploy before merging pull request
- Debug flaky test from standup chat
- Check cross-browser results after hotfix
- Review nightly test run failures
- Trigger smoke tests on staging environment
Integration
- Vendor
- Bugbug
- Category
- developer-tools
- Auth
- API_KEY
- Tools
- 4
- Composio slug
bugbug
Tools
- Get test run details
Get detailed information about test runs from bugbug. this action retrieves test run details with optional filtering parameters.
- List suites
List all available test suites in the bugbug platform. this tool makes a get request to the /suites/ endpoint and returns a list of available test suites with pagination support.
- List tests
Retrieves a list of all available tests in the bugbug account. this action provides access to all tests in the account with pagination support. it returns basic information about each test including its id, name, description, and timestamps
- Run test
Execute a test in bugbug and return the test run details. this action allows you to run a specific test with optional configuration for browser, device, viewport, and variable overrides. it returns detailed information about the test run in
Setup
Setup guide
- 11. Open your Switchy workspace and navigate to Settings > Integrations. 2. Find Bugbug in the developer tools section and click Connect. 3. Log into your Bugbug account and go to Settings > API Keys. 4. Generate a new API key (or copy an existing one) and paste it into the Switchy connection dialog. 5. Click Save to authorize the integration. 6. Return to any Space in Switchy and type '@Bugbug list suites' to confirm the connection works. 7. If you see your test suites appear, the setup is complete. 8. To run a test, use '@Bugbug run test' followed by the test ID and any browser or viewport overrides you need. 9. To review a past run, use '@Bugbug get test run details' with the run ID from your Bugbug dashboard.
What teammates see: by default, memories from Bugbug are scoped to the Space (PROJECT visibility) - you can mark any memory PRIVATE or share it ORG-wide.
Works well with
Top models
Compatibility data appears once enough Spaces have used this MCP together with a given model.
How Switchy teams use it
Starter prompts
List All Test Suites
@Bugbug list all available test suites in our account. Show me the suite names and IDs.Open in a Space →
Run Smoke Tests
@Bugbug run the smoke test suite using Chrome browser on desktop viewport. Let me know when the run finishes and if any tests failed.Open in a Space →
Check Last Run Status
@Bugbug get details for the most recent test run. Include pass/fail counts and any error messages from failed tests.Open in a Space →
Run Mobile Checkout Test
@Bugbug run the checkout flow test with mobile viewport settings. Override the test environment variable to use staging.Open in a Space →
List All Available Tests
@Bugbug list all tests in our account. Show me the test names, IDs, and which suite each belongs to.Open in a Space →
Example outputs
Illustrative - representative of the model's voice and quality, not literal recordings.
@bugbug list all test suites in our account and show me which ones have the most tests
Found 8 test suites in your Bugbug account. The 'Checkout Flow' suite contains 23 tests (most active), followed by 'User Authentication' with 17 tests and 'Product Search' with 12 tests. The remaining suites ('Dashboard', 'Settings', 'API Integration', 'Mobile Responsive', 'Admin Panel') each contain between 4-9 tests. All suites are currently enabled and have run within the last 7 days.
This example shows the MCP's read capability for test organization. Requires a valid API key with read access to suites. The count and suite names will vary by your Bugbug account setup — this illustrates the aggregation pattern, not your actual data.
@bugbug run the 'Login with Google OAuth' test on Chrome desktop and tell me if it passed
Initiated test run for 'Login with Google OAuth' (test ID: tst_9k2m4n). Running on Chrome 120, desktop viewport 1920x1080. Test completed in 18 seconds with status: PASSED. All 7 steps executed successfully, including OAuth redirect, token exchange, and profile data retrieval. Screenshots and video recording are available in the Bugbug dashboard. No console errors detected.
This demonstrates the MCP's ability to trigger test execution and report results. The test runs asynchronously — response time depends on test complexity (typically 10-60 seconds). Running tests consumes your Bugbug plan's monthly execution quota, so be mindful of usage in automated workflows.
@bugbug get the last 5 test runs for our checkout suite and summarize any failures with their error messages
Retrieved 5 most recent runs for 'Checkout Flow' suite (last 24 hours). 4 runs passed, 1 failed. The failed run (run ID: run_7h3p2q, 6 hours ago) encountered an error at step 'Apply Discount Code': element selector '.promo-input' timed out after 10 seconds. This suggests either a UI change broke the test or the promo code feature was temporarily unavailable. The failure occurred on Firefox 121, macOS. Previous runs on Chrome passed without issue.
This example pairs the MCP's test run retrieval with AI analysis to surface actionable insights. Useful for daily standups or incident triage. The MCP returns structured run data; the AI interprets patterns and failure context. Filtering by suite or date range helps manage large test histories.
Use-case deep-dives
When Bugbug MCP beats manual QA checks after staging pushes
A 6-person dev team ships to staging 3-4 times a day and needs someone to confirm critical flows still work before merging to prod. The Bugbug MCP is the right call here because you can wire a single Switchy prompt to run your core suite (login, checkout, admin panel) and parse the results in under 90 seconds. The AI reads the test run details, flags failures with stack traces, and posts a summary to Slack—no one opens the Bugbug UI unless something breaks. This works cleanly up to about 15 tests per suite; beyond that, the run-time and result-parsing get slow enough that you want a dedicated CI step instead. If your staging cadence is daily or slower, skip the MCP and just check Bugbug manually.
Why this MCP helps support triage visual regression reports
A 3-person support team gets tickets like 'the button is broken on mobile Safari' and needs to confirm the issue before escalating to engineering. The Bugbug MCP lets them ask Switchy to run the relevant test with device and viewport overrides, then read the result without learning Bugbug's API or waiting for a dev to reproduce it. The MCP's run-test tool accepts browser and device parameters, so the AI can match the customer's environment in one shot. This is faster than manual testing but only makes sense if you already have Bugbug tests covering the reported flows—if you're writing tests on-demand for each ticket, the overhead kills the speed gain. Best fit: teams with 20+ maintained tests and support staff who don't want Bugbug logins.
When the MCP saves time on cross-environment regression audits
A solo founder or 2-person team runs end-to-end tests weekly across staging, prod, and a preview branch, then reviews failures to decide what to fix. The Bugbug MCP is a good fit because you can ask Switchy to list all suites, pull the last 7 days of runs, and summarize which environments are clean versus which have recurring failures. The AI handles the pagination and filtering so you don't click through Bugbug's UI or write a script. The trade-off: if you run tests daily or more, the result volume makes the MCP's list-and-filter approach too slow—you want webhooks or a dashboard at that cadence. This works best for low-frequency, high-context review where you need a narrative summary more than raw data.
Frequently asked
What does the Bugbug MCP do in Switchy?
It lets AI agents trigger browser tests, list test suites, and retrieve test run results from your Bugbug account. Agents can run specific tests with custom browser or viewport settings, then pull the results to debug failures or verify deployments. Useful for automated QA workflows where the AI needs to confirm a feature works before marking a task complete.
Do I need admin access to connect Bugbug?
You need a Bugbug API key with read and execute permissions. Any team member with API key generation rights in Bugbug can connect it. The key lives in Switchy's credential store, so you don't need to be a Bugbug admin—just someone who can create an API key in your Bugbug workspace settings.
Can the MCP create or edit Bugbug tests?
No. It can only run existing tests, list suites, and fetch results. If you need to modify test steps or create new tests, do that in the Bugbug UI. The MCP is read-and-execute only—it won't change your test definitions or suite configurations.
Why use this instead of Bugbug's API directly?
The MCP wraps Bugbug's API so AI agents can trigger tests and parse results without you writing integration code. If you're already calling Bugbug from CI scripts, keep doing that. Use the MCP when you want an agent to decide which tests to run based on conversation context or pull results into a Switchy thread.
Does running tests through Switchy count against my Bugbug plan?
Yes. Every test run triggered via the MCP consumes a test execution from your Bugbug plan, exactly as if you ran it in the Bugbug dashboard or via their API. Switchy doesn't add extra usage—it just routes the request through your existing Bugbug account limits.