Agent Hub API
Agent Hub is the execution engine of VDF AI. It holds the agent registry and the tool catalogue — built-in tools, HTTP tools you define, and the remote MCP servers you register — and it runs agents against them.
Two concepts shape most requests. Workspaces, called Projects in the VDF AI portal, group the agents, tools, knowledge sources, and data connections that belong together. Domains group agents by subject, so a workspace can include a whole family of agents at once. Most list endpoints accept a workspace filter, and when you run an agent you can name the workspace it should work in.
Agent Hub also serves its tool catalogue to external MCP clients through the MCP endpoint.
Resources
Create, inspect, update, and delete agents, and run one against a prompt.
Project-style containers that scope agents, tools, knowledge, and connections.
Subject groupings over agents, with a model portfolio and per-user tool overrides.
Browse the tool catalogue, run a tool directly, and manage your own HTTP tools.
Register remote Model Context Protocol servers so that agents can use the tools they expose.
A Model Context Protocol server that exposes your tool catalogue to MCP-compatible clients.
Create, share and version reusable instruction bundles that agents load when a task matches.
List the language models in your deployment's model catalogue and look up their capabilities.
Review your agent runs: prompts, outputs, token usage, status and timing.
List, read, rename and delete your agent conversations and the messages in them.
List, download and delete the files that agents generate for you, such as reports, spreadsheets, charts and images.
Give colleagues or permission groups access to skills, tools, MCP servers and agents that you own.
Keep an EU AI Act register of your AI systems, scan their repositories, and track the risks and remediation actions that follow.
AdministrationEndpoints for the administrators of your deployment.
Read and change the Agent Hub runtime configuration, and read its recent error log.
OperationsHealth checks for monitoring and load balancers.
Liveness probe that reports whether the Agent Hub service is running.