API reference
The VDF AI API is organised around REST. It has predictable, resource-oriented URLs, accepts JSON request bodies (and multipart form data for file uploads), returns JSON responses, and uses standard HTTP methods, status codes, and bearer authentication.
VDF AI runs on your infrastructure, so the API is served by your own deployment, on the same domain your people already use to open VDF AI. There is no separate API host and no VDF-operated service in the path of your data.
This reference covers the endpoints you use to integrate applications with VDF AI: identity and sessions, conversations and documents, agents and the tools they use, connected systems and the knowledge indexed from them, and governed data.
# The HTTPS origin of your VDF AI deployment
export VDF_BASE_URL="https://vdf.example.com"
# An access token from Sign in (see Authentication)
export VDF_ACCESS_TOKEN="<ACCESS_TOKEN>" Explore the APIs
/api Chat API Sign-in, conversations, documents, and the integrations that feed your knowledge base. /agent-hub-api Agent Hub API Agents, workspaces, tools, MCP servers, and skills — and the endpoint that runs them. /networks-api Networks API Design and run multi-agent networks, follow every run, and turn intent into a network. /data-api Data API Governed database connections, profiling, feature discovery, fine-tuning datasets, and vector search. Base URL and services
Every API is reachable through your deployment's gateway under its own path prefix, so one hostname and one TLS certificate cover the whole platform. Append the paths in this reference to the HTTPS origin of your deployment.
| API | Path prefix | Covers |
|---|---|---|
| Chat API | /api |
Sign-in, conversations, documents, and the integrations that feed your knowledge base. |
| Agent Hub API | /agent-hub-api |
Agents, workspaces, tools, MCP servers, and skills — and the endpoint that runs them. |
| Networks API | /networks-api |
Design and run multi-agent networks, follow every run, and turn intent into a network. |
| Data API | /data-api |
Governed database connections, profiling, feature discovery, fine-tuning datasets, and vector search. |
Each endpoint in this reference is shown with its full path, prefix included.
https://vdf.example.com/agent-hub-api/api/agent/list
└──────── your deployment ─────┘└── prefix ──┘└─ path ──┘ Your first request
Sign in with a VDF AI user account to obtain an access token, then send that token as a bearer token on every request. The example on the right stores the token in VDF_ACCESS_TOKEN and lists the agents you can use.
Every example in this reference reads two environment variables: VDF_BASE_URL, the HTTPS origin of your deployment, and VDF_ACCESS_TOKEN. The examples are illustrative and inert — this site never sends requests on your behalf and never asks for your credentials.
export VDF_ACCESS_TOKEN="$(
curl -s -X POST "$VDF_BASE_URL/api/auth/login" \
-H "Content-Type: application/json" \
-d '{"email": "ada@example.com", "password": "<PASSWORD>"}' \
| jq -r '.access_token'
)"
curl "$VDF_BASE_URL/agent-hub-api/api/agent/list" \
-H "Authorization: Bearer $VDF_ACCESS_TOKEN" Releases and versioning
The API is versioned with the platform. Your deployment serves the API of the VDF AI release installed on it, and upgrading your deployment upgrades its API. This reference tracks the current release; if an endpoint described here is not available on your deployment, your installation predates it.
Review the release notes before you upgrade a deployment that integrations depend on.
What this reference covers
This reference documents the endpoints for building integrations — everything an application, a script, or an automation needs to work with VDF AI on behalf of a signed-in user — and the administration endpoints your deployment's administrators use, such as user and group management, single sign-on, backups, and platform settings. Administration pages are grouped separately under each API and marked on every endpoint.
The Model Catalog and evaluation tooling are documented in the API reference inside the VDF AI portal of your deployment, as are internal service-to-service interfaces.