Skip to main content
Building with an AI agent? Point it at these and it can integrate on its own:Create a key on the API Keys page, authenticate with Authorization: Bearer YOUR_API_KEY, then hand the agent the OpenAPI spec and let it go.
Or just hand your agent the prompt (Cursor, Claude Code, Codex, OpenCode, or harness of choice):

​
Explore the full API

The complete API is described by an OpenAPI 3.1 spec. Browse it interactively or hand the raw spec to your client or agent.

OpenAPI spec

The full machine-readable spec (openapi.json). Import it into your client or pass it to an agent.

Swagger UI

Navigate and try every endpoint interactively.

​
Base URL

The Invent API is built on REST principles. We enforce HTTPS in every request to improve data security, integrity, and privacy. The API does not support HTTP. All requests contain the following base URL:

​
Authentication

To authenticate, you need to add an Authorization header with the contents being Bearer YOUR_API_KEY.
Go to the API Keys page and create a new API key.

​
Organization-scoped routes

Resources are under /orgs/{org_id}/... (for example GET /orgs/{org_id}/assistants).
  • {org_id} = c: with Bearer API key auth, c is always the org the key was issued for (e.g. the parent if the key was created on the parent). Example: GET https://api.useinvent.com/orgs/c/assistants.
  • Explicit org_id: use the real organization id for any other workspace, including a sub-organization when the key is a parent key. c does not refer to a sub-org, so you must pass the sub-org’s id in the path. See sub-organizations in API Keys.

​
Response codes

Invent uses standard HTTP codes to indicate the success or failure of your requests. In general, 2xx HTTP codes correspond to success, 4xx codes are for user-related failures, and 5xx codes are for infrastructure issues.

​
Rate limits

Invent enforces rate limits to ensure fair usage of the API. If you exceed a rate limit, you will receive a 429 status code. Limits are applied per endpoint and keyed per caller (your API key’s owner, or IP address). They vary by action: write-heavy endpoints are capped much tighter than reads — for example, creating an assistant is limited to 10 per hour. Expect 429s on those well below any general per-minute ceiling, and back off when you receive one. A broad edge limit of roughly 500 requests per minute per IP may also apply at the gateway.