Skip to main content
Henry is built to be used from chat, not programmed against. Most teams never need this guide. It exists for the cases where you want to plug your own systems in, or need precise answers about how Henry handles your data.
Henry does not offer a general-purpose REST API with issued API keys today. The endpoints described here are specific and narrow: an MCP endpoint for coding agents in CI, and cookie-authenticated downloads for exports. If you need programmatic access to something else, email support@usehenry.ai.

What you can do

Custom MCP servers

Give Henry your internal tools by pointing it at any Streamable HTTP MCP server. OAuth, static header, or no auth.

Bring your own model key

Enterprise workspaces can run Henry on their own Anthropic or OpenAI account.

Coding agent access

Let coding agents in your CI read your deployments and logs through Henry with a scoped, read-only token.

Permissions Henry requests

The exact Slack scopes and Microsoft Teams permissions, and why each one is there.

Exports

Workspace archive layout, audit log CSV columns, and usage CSV grains.

Security architecture

Encryption, tenant isolation, prompt-injection posture, sandboxing, and retention windows.

Self-hosted

Running Henry in your own environment against your own models.

Endpoints at a glance

The one URL you may need to register on your side is the OAuth redirect for custom MCP servers: https://app.usehenry.ai/api/integrations/custom-connectors/callback.