Usage guide
Check the use case, inputs, and reading focus first, then decide whether this method fits your question.
The API page serves people who want to connect their own LLM or agent to the site analysis workflow. Its job is not just to show an API key. It should explain how to choose a method, preserve inputs, tell the user the current credit balance and the cost of the selected method, and redirect to Pricing when credits are insufficient.
If you want to connect the site to your own agent, this page explains the single analysis workflow, how credits work, whether reports are saved to Profile, and how to choose methods more reliably.
These sections add more context for how this method is usually used in practice.
The product path is now focused on delivering the same finished reports as the site itself. A single analysis workflow keeps credit rules, Profile persistence, predictive combined-report behavior, and reporting consistency aligned with the main product.
A reliable agent integration should not wait for a failed charge to mention credits. It should read the current account balance first, read the selected method cost from the method definition, and explicitly tell the user how many credits remain and how many this analysis will consume.
When the analysis job returns an insufficient-credit error, the agent should immediately explain the remaining balance, the required cost, and send the user to the Pricing page to purchase more credits before continuing the current flow.
The API covers every divination method available on the site. Your agent can first use the recommendation endpoint, then inspect required fields and credit cost for the selected method, and finally create an analysis job.
Yes. The Agent API now uses the site analysis workflow, so each request spends credits according to the method definition. Most single methods cost 1 credit, while some deeper or combined-report products cost 2 credits.
Yes. Agent API analysis jobs create normal Profile reports, and those reports are automatically labeled as API so they can be filtered separately.
The analysis job should return an insufficient-credit error together with the remaining balance, required credits, and a Pricing URL. The external agent should then send the user to Pricing, let them purchase credits, and continue the flow afterward.
If you want to compare methods, question styles, or usage boundaries further, these are good next topics.