TestMyVibes MCP server

Hosted Streamable-HTTP MCP server. Connect any MCP-aware client (Claude Code, Cursor, ChatGPT, Codex, Continue, Cline) and your coding agent can submit AI-driven QA tests against any URL — fix-test-retest in a closed loop.

Endpoint

https://testmyvibes.com/mcp

Transport: streamable-http (also SSE-compatible). Authentication: Authorization: Bearer <tmv_live_*>. Get a key by signing up at https://testmyvibes.com/auth/register; smallest credit pack is $5 (50 credits at $0.10 each).

Browse without auth: initialize, tools/list, resources/list, and prompts/list are callable anonymously so reviewers can audit the catalog before signing up.

Quick install

Claude Code

claude mcp add testmyvibes \
  --transport http \
  --url https://testmyvibes.com/mcp \
  --header "Authorization: Bearer $TMV_API_KEY"

Cursor / Continue

Add to .cursor/mcp.json or ~/.continue/config.json:

{
  "mcpServers": {
    "testmyvibes": {
      "url": "https://testmyvibes.com/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_TMV_API_KEY"
      }
    }
  }
}

Claude Desktop

Edit claude_desktop_config.json:

{
  "mcpServers": {
    "testmyvibes": {
      "command": "npx",
      "args": ["mcp-remote", "https://testmyvibes.com/mcp", "--header", "Authorization: Bearer YOUR_TMV_API_KEY"]
    }
  }
}

What your agent can do

Full MCP tool catalog → (lives + generated from code).

Headline tools:

• submit_test — queue an AI browser test against any URL. 11 personalities, fixed pricing, refund-on-underrun.
• submit_combo — packaged multi-agent bundles. Smoke Stack → Whole Kit & Kaboodle (every agent × multiple pages). Auto-pauses + refunds when too many bugs found.
• submit_interaction_scene — multi-agent coordinated scenes (publisher+viewer, buyer+seller, multi-user chat) with signal/wait_for_signal primitives.
• capture_screenshots — on-demand viewport captures (mobile + tablet + desktop bundle).
• provision_test_card — Stripe Issuing virtual cards for real-checkout testing (coming soon — Stripe activation pending).
• list_kept_personas + existingPersonaId — kept-alive personas for return-user testing.
• retest_job — re-run the same test post-fix without restating parameters.
• get_test_results — fetch full diagnostic envelope: action trail, console errors, failed network requests, screenshots, AI summary.

The fix-test-retest loop in 4 calls

1. submit_test({ url, description, goal, agentInstructions, useTestInbox: true })
   → { jobId }

2. get_test_status({ jobId })   // poll every ~30s

3. get_test_results({ jobId })  // diagnostic envelope: actionTrail,
                                //   consoleErrors, failedRequests,
                                //   screenshots, bugs[]

4. 

5. retest_job({ jobId })        // re-runs same test against latest
                                // deployment — preserves URL, goal,
                                // inbox config
   → loop back to 2

Pricing

1 credit = $0.10 USD. Smallest credit pack is $5 (50 credits). Per-tool pricing on the /pricing page (live from code, never drifts). Volume discount kicks in at the $450 pack.

Sample costs (at $0.10/credit):

• Cheapest AI smoke check: ~4 credits / $0.40 / 2 min
• Signup gauntlet with email OTP: ~10 credits / $1 / 5 min
• Marketing CRO audit (deep): ~18 credits / $1.80 / 6 min
• Whole Kit Solo (1 story × 13 agents): 216 credits / $21.60 / 60 min, auto-pauses if site is broken
• Whole Kit Core (3 stories): 580 credits / $58 / 120 min
• Whole Kit Plus (6 stories): 1080 credits / $108 / 180 min
• Whole Kit Max (10 stories): 1890 credits / $189 / 240 min

For agents bringing your own agent

Tell your coding agent: WebFetch https://testmyvibes.com/docs/coding-agents — the longer integration guide explains the test-retest loop, agentInstructions style, when to pick AI vs human runner, and the diagnostic envelope shape.

Support

support@testmyvibes.com · Support center