TrueFetch documentation

Collect structured job records with explicit source outcomes and credit budgets. Choose the path that fits your first task, then retain its reference for results and recovery.

Console quickstartSign in, inspect a quote and explicitly start your first task.

REST quickstartQuote, submit and recover a task with an authorized API key.

Agent and workflow examplesChoose a downloadable framework kit with host budgets and original-command recovery.

Job Search APIInputs, source support and limits for the available Job Search API.

Response outcomesSeparate completed, partial and failed collection.

MCP planPlanned tools, authentication and client access.

Walk through the Job Search example

1. Inspect the product sample without a key or charge. The homepage also shows a dated real retained output; fictional outcome fixtures below are labeled separately.
2. Check the product scope and fields before choosing sources.
3. Interpret outcomes, including partial delivery and a successful empty search.
4. Understand the reservation and settlement before authorizing spending.
5. Read authentication, pagination and retries and error recovery.
6. Check release availability before connecting an API or an AI client.

What is available now?

Capability	Availability
Accounts & workspace	Available
API execution	Available
Credit purchases & paid usage	Available
MCP connections	Not open

This is release availability, not a live health check. View release status.

Job Search, its REST quickstart, and the account workspace provide the current collection path. Planned products and remote MCP access remain labeled by release availability. Framework kits have their own preview qualifications; local fake-transport checks do not establish live-provider acceptance. See limits and source support, costs, results and retention, or contact support when choosing your first task.

Build workflows and connect agents

For code, use the executable requests in the REST quickstart. For agents and automation, choose an integration example; these kits use the canonical REST routes with host-controlled budgets. The Claude Agent SDK kit exposes an in-process SDK server, separate from TrueFetch’s remote MCP release. For remote AI clients, read the MCP plan and current availability before attempting a connection.

Naming

Public JSON fields use camelCase. API and planned MCP operations share the product contract. Source-specific support is explicit; a source cannot silently ignore a requested filter.

Download Job Search sample responses

Fixed fictional fixtures checked against the contract on September 24, 2026. These files do not query sources or consume credits. The four task files below contain { "task": ... }; records come from a separate result request. Times and billing amounts are illustrative.

• Succeeded task: three records
• Partial task: one source unavailable
• Empty task: a completed search with no matches
• Failed task: no sources completed

Separate result responses:

• Succeeded result page
• Partial result page
• Empty result page
• Failed result request: HTTP 409

Inspect the failed task before retrying: a retry hint on the result endpoint does not restart a terminal task. A new collection needs a new request key and budget. See tasks and results for pagination and JSONL/CSV downloads.