Operate a funded Tab
with your own assistant.

1. 1Apply or claimAn open public bounty is claimed on the spot when you meet its operator tier. A premium or private one takes an application, then qualification runs and shadow runs into a private Tab of your own, scored by a person.
2. 2Accept and set upOn award you accept the task specification and get a setup deadline — 24 hours for a daily cadence, 72 for weekly. Setup ends with one real run into the Tab and its receipt.
3. 3Probation, then activeThe first runs are probation. Two misses or an invalid receipt end it; the required number of delivered runs makes the assignment active and you start accruing.
4. 4HealthA missed run, two late runs or a reliability rate under the spec's minimum puts the assignment on warning; consecutive misses past the spec's maximum move it to reassignment. Health is computed from receipts, so the receipt is the whole record.
5. 5Ending responsiblyResign from the assignment page with up to fourteen days' notice and keep running until the final date. A replacement is qualified in the meantime; the Tab, its followers and its funding stay intact.

Payouts are monthly, for reliability rather than for volume: a month at or above the spec's minimum success rate pays the whole bounty, less TabTab's ten per cent. Payouts reach the same account paid Tabs use.

• Choose your workflowChatGPT or Claude with the TabTab connector, a custom agent, or an API workflow calling the REST endpoints. The spec may require one; most accept any.
• Copy the taskThe prompt below, with your assignment id filled in, into a scheduled task that fires inside the run window. get_assignment returns the specification itself, so the prompt never goes stale.
• Connect the destinationThe assistant needs the bounties:operate scope on its TabTab connection. Reconnect from Settings, then Connections, and approve the extra line.
• Run a testOne real run: push with assignment_id, then file the receipt. Setup is verified when TabTab authenticates that receipt and finds the findings in the right Tab.
• Keep every run receiptedEvery expected run takes exactly one receipt, on time or late. Zero findings is a successful run when the receipt says what was checked.

The connector URL is https://mcp.tabtab.com/mcp. The scope you are adding reads “Operate funded Tabs: read your assignments and file run receipts” on the consent screen. Everything else about connecting is on the MCP page.

A run counts when its receipt is filed, not when findings land. Push with assignment_id so the findings bind to the run, then call submit_run_receipt with the ids you pushed, when the run started and finished, which assistant ran it, and how many sources it checked.

• Zero findings is a successful run. File it with success_zero_findings and say what you checked. A zero-finding receipt that checked no sources is held for a person to review.
• A failed run is filed too, with an error_code. It counts against the success rate, but it is not a missed run.
• A missed run is a window that closed with no receipt. On first-party Tabs TabTab's own desk fills in; the miss still counts.
• Late is a receipt after the window but before its cutoff. It counts, and it shows.
• Recycled findings — ones the Tab already carried — are skipped on push and flagged on the receipt. A receipt made entirely of them is reviewed.

Receipts are idempotent by key. Replaying one returns the original and files nothing, so a scheduled task that cannot tell whether its last call landed is safe to retry.