GetABrain.ai

Give your AI agent access to real human judgment with our Human-in-the-Loop MCP Server.

CommunityOnlineAPI key requiredGlobalFreeCan modify data

What it can do

  • Get Balance: Check the prepaid balance (in cents) available to fund submit_query calls, plus account mode. Call before submit_query if unsure funds suffice, or whenever a query fails/stalls for balanc
  • Create Topup Link: Generate a one-time Stripe Checkout URL for adding funds to the prepaid balance. Use when get_balance shows insufficient funds for an upcoming submit_query, or a human asks to add m
  • Submit Query: Submit a structured question to real human workers, returning a query_id -- the entry point for any human-in-the-loop (HITL) step: judgment calls, subjective evaluation, approval/review,

What data it sees

Do you need an account

An API key from the service settings is required

Give your AI agent access to real human judgment with our Human-in-the-Loop MCP Server.

When your agent encounters a question it cannot answer reliably, it sends the request to a real pool of humans and receives a structured response back.

Ideal for:

  • Judgment calls and subjective ratings
  • Real-world and local knowledge
  • Image and audio review
  • Quality checks and human verification
  • Tasks where accuracy matters more than guessing

Includes 7 MCP tools, a free sandbox mode for testing, and simple pay-per-answer pricing when you go live.

Works with Claude Desktop, Cursor, and any MCP-compatible client.

How to connect

How to connect GetABrain.ai to Claude

The server address to paste into Claude settings:

https://getabrain.run.tools

  1. Open Claude (claude.ai or the desktop app).
  2. Go to Settings → Connectors.
  3. Click Add custom connector.
  4. Paste the server address copied below into Remote MCP server URL and click Add.
  5. In a chat, click + → Connectors and switch the new connector on.

Custom connectors are available on Free, Pro, Max, Team and Enterprise plans (Free is limited to one). On Team and Enterprise an organization Owner adds the connector first under Organization settings → Connectors.

Authorization

After you click Add, a sign-in window for the service opens. Sign in with your own account and approve access. Claude never sees your password.

How to connect GetABrain.ai to ChatGPT

The server address to paste into ChatGPT settings:

https://getabrain.run.tools

  1. Open ChatGPT in a browser (chatgpt.com). A Plus, Pro, Business, Enterprise or Edu plan is required.
  2. Turn on developer mode once: Settings → Apps → Advanced settings → Developer mode.
  3. Open Settings → Connectors and click Create.
  4. Fill in the form: Name (anything), Description (one line about what the service does), MCP server URL (copy it below).
  5. Under Authentication choose OAuth if the service requires sign-in, otherwise None. Click Create.
  6. In a new chat open + → Apps/Connectors and enable the connector.

OpenAI has renamed this section before (Connectors → Apps/Plugins). If the label differs, search settings for "developer mode". On Business/Enterprise workspaces an admin must allow custom connectors first.

Authorization

On first use ChatGPT opens the service's sign-in window. Sign in and approve access.

How to connect GetABrain.ai to Cursor

The server address to paste into Cursor settings:

https://getabrain.run.tools

Fastest: click Open in Cursor below and confirm the prompt.

Manually:

  1. In Cursor open Settings → Cursor Settings → MCP and click Add new global MCP server.
  2. Paste the JSON copied below into ~/.cursor/mcp.json (per project: .cursor/mcp.json in the repo root).
  3. Save the file. The server appears in the MCP list; authorize it there if asked.

Authorization

If the service needs sign-in, an authorize button appears next to the server in the MCP list.

Server tool list (7)

Raw names from tools/list. Only developers need these.

get_balanceCheck the prepaid balance (in cents) available to fund submit_query calls, plus account mode. Call before submit_query if unsure funds suffice, or whenever a query fails/stalls for balance reasons. Read-only, no side effects, free in test and live mode. Response includes mode ("test" = sandbox key, free simulated responses; "live" = real key, real spend), and when relevant whether auto_reload is on, plus an auto_reload_setup_url + hint to enable it so a live account does not stall at zero balance. Disambiguation: reports funds available to spend; does not list queries (list_queries) or responses (get_responses/wait_for_responses).
create_topup_linkGenerate a one-time Stripe Checkout URL for adding funds to the prepaid balance. Use when get_balance shows insufficient funds for an upcoming submit_query, or a human asks to add money. Side effect: creates a pending Stripe session (no charge yet); the returned checkout_url must be OPENED IN A BROWSER BY A HUMAN to enter payment details and complete the charge -- the agent cannot complete payment itself; this is an out-of-band, human-in-the-loop step. Works with a test-mode or live-mode key (funding a test-mode account is how you move a sandbox integration to real spending power). Balance updates only after checkout completes; poll get_balance to confirm. Disambiguation: only mints a payment link -- never moves money or blocks waiting for payment itself.
submit_querySubmit a structured question to real human workers, returning a query_id -- the entry point for any human-in-the-loop (HITL) step: judgment calls, subjective evaluation, approval/review, or "ask a real person" tasks a model should not answer itself. "type" selects the question format (one of: text, multiple_choice, rating_scale, image_comparison, ranking, yes_no, sentiment, image_selection, free_form_text, video_review, audio_review, image_analysis, ab_test, voice_capture, video_capture, photo_capture); "content_data" is the matching type-specific payload (e.g. ab_test: {question, variant_a:{description}, variant_b:{description}}; yes_no: {question}; rating_scale: {question, scale_type, scale_min, scale_max}). Cost/side effects: with a LIVE key this deducts (bid_amount_cents + bonus_amount_cents) * required_responses from balance immediately and dispatches to paid workers (fails if balance too low -- check get_balance or use create_topup_link). With a TEST key, no balance is touched and responses are synthetic, marked simulated: true, so you can build/test a full pipeline for free before going live. Returns immediately, does not wait -- use get_responses (one-shot) or wait_for_responses (bounded polling) to retrieve answers.
get_responsesOne-shot read: fetch a query's current status and whatever human responses have been submitted so far, without waiting. Use this to check progress on demand, or after wait_for_responses reports "pending" if you want an immediate snapshot instead of polling again. Read-only, no cost, returns instantly (does not block or retry). Disambiguation: unlike wait_for_responses, this never delays or blocks waiting for more answers to arrive -- it just reports what exists right now, which may be fewer than required_responses.
wait_for_responsesPoll for human responses to a query, blocking for up to max_wait_seconds (default 50, max 50) before returning. Use right after submit_query to wait for real answers in one call instead of manually re-checking with get_responses. DOES NOT GUARANTEE COMPLETION -- if min_responses have not arrived within the time budget it returns status "pending" (with a hint to call again) rather than erroring; call again to keep waiting. Returns status "ready" with the responses array once enough have arrived (or the query otherwise completed). Read-only / free -- cost was already charged by submit_query. Disambiguation: unlike get_responses (instant, one-shot, may return 0 responses), this actively waits, trading time for a higher chance of a complete result.
list_queriesList your recent GetABrain queries, most recent first. Use this to get an overview of past/active queries, recover a query_id you lost track of, or filter by status (e.g. find everything still "active" or "pending"). Read-only, no cost. Disambiguation: this lists MANY queries at a summary level; it does not return the individual worker responses for any one query -- use get_responses or wait_for_responses with a specific query_id for that.
rate_responseRate a single worker's response 1-5 to feed the worker quality/reputation system, optionally with free-text feedback. Use after reviewing a response from get_responses/wait_for_responses, to reward good answers and flag poor ones -- this affects the worker's quality score and future eligibility (e.g. queries with min_worker_quality set) and can trigger rewards/suspension server-side. Side effect: writes a rating record and returns the updated worker quality score; does not resubmit or modify the original response. Disambiguation: rates a response you already have -- does not fetch new responses (use get_responses/wait_for_responses first).