Skip to main content
OpenConnector exposes Eden AI as two actions: a model list and a non-streaming chat completion. This page runs the gateway with Docker, stores your Eden AI key as a connection and calls the edenai service.

Overview

OpenConnector is an open-source, self-hosted auth gateway that connects SaaS providers to AI agents through SDK, CLI, MCP, HTTP and OpenAPI. Eden AI is one of its built-in connectors, so an agent wired through OpenConnector can call models from OpenAI, Anthropic, Google, Mistral and more behind one key, with the Eden AI billed cost when Eden AI returns one. The connector exposes chat completions (non-streaming only) and a model list. For EU data residency, see EU Data Residency Endpoint: the connector’s base URL is fixed to https://api.edenai.run/v3 and cannot be pointed at the EU endpoint.

Prerequisites

Setup

1. Start OpenConnector

Clone the repository and start the published image:
This pulls ghcr.io/oomol-lab/open-connector:latest and serves the console at http://localhost:3000 and the API reference at http://localhost:3000/docs. For anything beyond a trial, pin a released version (for example v1.9.4) instead of latest.

2. Store your Eden AI key

OpenConnector checks the key with a request to Eden AI’s /models, so a bad key is rejected at this step. If you set OOMOL_CONNECT_ADMIN_TOKEN, send it as authorization: Bearer <token> on /api calls. The connector reads no Eden AI environment variable. The key is stored as the edenai connection (authType: api_key, field apiKey) and sent to Eden AI as a bearer token. Run the same PUT with a new key to replace it. Set OOMOL_CONNECT_ENCRYPTION_KEY to encrypt stored credentials at rest.

3. Call Eden AI

If you created a runtime token, send it as a bearer token on /v1 calls. The response includes the standard OpenAI-compatible fields plus a top-level cost field with the Eden AI billed cost, when Eden AI returns one. stream: true is rejected.

Model naming

The edenai service exposes two actions: edenai.list_models and edenai.create_chat_completion. Use the provider/model format, for example openai/gpt-latest, anthropic/claude-sonnet-latest or mistral/mistral-large-latest. Call edenai.list_models or see List LLM Models for the catalog.

Troubleshooting

401 from /api or /v1

An admin token or runtime token is configured. Send it as a bearer token.

The key is rejected when you store it

OpenConnector validates the key against Eden AI’s /models. Eden AI answers 401 with Invalid API key for a wrong or revoked sk-eden- key and Invalid token for other token shapes. Create a regular API key at app.edenai.run.

Streaming fails

The connector accepts only stream: false or no stream field.

Next Steps