Felo API PlatformFelo API Platform

AI Tool Setup

AI Tool Setup Overview

Connect your AI tools to Felo: prepare an API key, choose a protocol and model, and follow a client setup guide.

Connect your coding tools, editors, and agents to models available through Felo API Platform. Your client manages the conversation, local files, and tool execution; Felo provides the model API. Configure an API key, a base URL, and a model ID to connect them.

Before You Start

  1. Create and save a key on the API Keys page.
  2. Confirm that your account has available service credits.
  3. Install a client that supports a custom model provider or compatible API endpoint.
  4. Choose a model from the catalog. Coding agents need tool-call support; copy the exact model ID.

Use your Felo key with a custom API provider. It is an API credential, separate from a client's own subscription login.

Choose What to Connect

  • Felo models: configure the client's model provider. Model calls use platform balance; see each client guide for supported CLI automation.
  • Felo Skill: add Felo tool capabilities to your existing model. For OpenCode, feloapi setup --agent opencode --skills-only --yes installs the Skill only; configure model access separately using the OpenCode guide.

See the FeloAPI CLI guide for installation, validation, and restoring client settings.

Choose Your Tool

WorkflowSetup guides
Terminal coding, code review, and repository changesCodex, Claude Code, OpenCode
Local assistants and agent workflowsOpenClaw, Hermes Agent
Editor model connectionsCursor, TRAE
Coding extensionsCline, Roo Code, Kilo Code
A local model gateway for multiple agentsMagpie
Other clients with a custom API endpointGeneral setup guide

Each guide covers provider settings, credentials, model selection, and verification. Custom provider support may depend on the client's version or account plan.

Choose the Protocol Endpoint

Client protocolBase URLTypical client
OpenAI-compatible Chat Completionshttps://openapi.felo.ai/api/v1OpenCode and extensions using this mode
OpenAI-compatible Responseshttps://openapi.felo.ai/api/v1Codex
Anthropic-compatible Messageshttps://openapi.felo.ai/apiClaude Code and clients using Messages

The client adapter adds paths such as /chat/completions, /responses, or /messages; do not append them to the base URL. Choose the protocol from the client's provider settings, not just the model name.

Connect and Verify

  1. Open your client's guide and select its custom provider or compatible API mode.
  2. Set the Felo key, protocol endpoint, and model ID. Prefer environment variables for credentials and reference them as required by the client.
  3. Select the Felo provider and model for the current session.
  4. Send a short message to verify a response.
  5. For coding agents, try a simple read-only task, such as listing the main project directories, to check tool calls.

Environment variable names and loading rules vary by client. Follow its guide rather than assuming every tool automatically reads FELO_API_KEY or OPENAI_BASE_URL.

Connection Problems

  • Authentication: check the complete, active key and the client's process environment.
  • Wrong service: check the selected provider and project settings that override user settings.
  • Endpoint errors: match the provider protocol to the base URL above.
  • Model or tool-call failures: check the model ID, availability, and tool support.

See Troubleshooting and Protocols and SDKs for more help.

Add More Capabilities

After connecting the model, see the Felo OpenAPI guide for web search, extraction, X data, and presentations. Configure these separately; changing the model provider does not enable them automatically. Check the relevant API guide for its billing rules.

Need help?