Using Your Agent

Four ways to put your Agentheya agent to work — chat link, REST API, MCP server, and scheduled runs.

Once your agent is configured, you have four ways to use it:

Method	Best for
Chat link	Sharing with users — no code required
REST API	Apps, bots, and workflows that speak the OpenAI API
MCP server	Claude Desktop, Cursor, Zed, and other AI coding tools
Schedules	Recurring tasks that run on a timer

All four share the same agent — the same configuration, tools, memory, and activity log. They differ only in how the conversation is initiated.

Chat link

A hosted URL your users open in a browser. No code required.

REST API

OpenAI-compatible. Works with any openai SDK, LangChain, and more.

MCP server

Call your agent as a tool from Claude Desktop, Cursor, Zed, and similar.

Schedules

Set a time or a cron expression. The agent runs itself.

API keys

The REST API and MCP server authenticate with the same API key format. Generate one from your agent’s Details page in the dashboard.

• Keys start with sk_ah_
• Each key is bound to one agent
• You see the full key once at creation — we only store a hash
• Both agent owners and subscribers can generate keys; subscriber keys scope access to that user’s own data

To expose multiple agents across multiple integrations, generate one key per agent and configure each separately.