> ## Documentation Index
> Fetch the complete documentation index at: https://docs.revcenter.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Agent-driven research

> Let an AI agent draft criteria, run a search and explain the results, with you approving each spend.

> Agents: the full documentation index is at [/llms.txt](https://docs.revcenter.ai/llms.txt).

An agent is good at the parts around a search: reading a brief, drafting criteria, explaining who fits and why,
and trying scoring changes. Revcenter does the searching and scoring, deterministically: the agent's criteria
are run exactly as sent (`compiled_by: "caller"`).

* **Spending is explicit.** Only launching a search (not `database_only`) and expanding use credits, 1 each,
  and every tool says so in its description. Tell the agent to ask before either.
* **Exploring is free.** Reading the criteria reference, people, previews of scoring changes and searches of your
  own candidates use no credits, so the agent can iterate.
* **Retries are safe** with an idempotency key; see [errors](/guides/errors#idempotency-keys).
* **The docs are built for agents**: [/llms.txt](https://docs.revcenter.ai/llms.txt) lists every page.

**Recipe:** [Run a search from a coding agent](/recipes/coding-agent). The full workflow is on
[Agents](/tools/agents).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.