---
title: "Xedant Agent Integration"
id: "1804"
type: "page"
slug: "agent"
published_at: "2026-09-20T21:51:54+00:00"
modified_at: "2026-09-21T01:36:28+00:00"
url: "https://xedant.com/agents/data/docs/agent"
markdown_url: "https://xedant.com/agents/data/docs/agent.md"
excerpt: "Data Agent runs on your central Xedant Agent: chat and every AI feature come from…"
---

# Xedant Agent Integration

[https://xedant.com/agents/data/docs/agent.md](https://xedant.com/agents/data/docs/agent.md)

Data Agent runs on your central Xedant Agent: chat and every AI feature come from it. No separate model keys and subscriptions are needed — Data Agent uses your agent’s models, skills and chat infrastructure, and concerns itself only with scripts, runs and data.

## Chat: the agent outside

The `AGENT_API_URL` and `AGENT_API_KEY` variables connect Data Agent to Xedant Agent. The application proxies the conversation: chat opens on every page, and the models and skills are yours, from the Xedant Agent panel. One agent works in Data Agent and in the other products alike — no per-app setup needed. More about chat is on the [Chat with the Agent](/agents/data/docs/chat)
 page.

## The shared scripts folder

Xedant Agent works on the same `scripts/` folder Data Agent mounts (it is the agent’s `AGENT_PROJECT_PATH` folder). Both sides see the same files: the agent writes and fixes the scripts, Data Agent runs and schedules them. The roles are separated and never conflict — the agent does not run scripts through the application; it only prepares them.

## Agent APIs: the agent inside

For data management Data Agent has programmatic interfaces — the `/api/agent/*` API endpoints, locked behind a key in the `X-API-Key` header (the `DATA_AGENT_API_KEY` variable stores this key’s SHA-256 hash). Through them the agent can:

- open ETL sessions tied to the chat;
- see the data sources and their schemas, and run queries (read-only: SELECT);
- work with the Lakehouse: listings, folders, files;
- manage scripts: list, run, poll status, stop;
- read and replace the schedule wholesale.

## The ETL cycle from chat

A typical cycle looks like this: the agent opens a session → looks at the available sources → writes or fixes `extract.py` and `load.py` → runs them → checks the data with a query → sets the schedule → closes the session. Runs made within the session carry its badge and a link to the chat — you always see what ran and why.

## API documentation

The agent APIs’ description lives in the Data Agent project (the `data-agent-api.md` file) — it is for developers and for the agent itself. As a user, it is enough to know that the agent can do all of this on its own: you simply describe the task in chat.

## License

Chat and prompts-queue management require a valid Data Agent license. License types and prices are on the [Data Agent Licensing](/agents/data/license)
 page.

More about Xedant Agent itself is on the [Xedant Agent](/agents/agent)
 page.

See also: [Chat with the Agent](/agents/data/docs/chat)
, [Scheduler](/agents/data/docs/scheduler)
.

[← Back to the documentation index](/agents/data/docs)