Integrations

› Agent engines

First-party coding agents plus anything custom you plug in. Traces, cost, and diffs follow every run.

Claude Code

Deep reasoning and refactors. Use your Anthropic subscription.

OpenAI Codex

Fast iteration. Bring your OpenAI API key.

Custom

Your own engine, CLI, or wrapper — wire it in and keep the same observability.

› Models

Anthropic Claude, OpenAI APIs, a ChatGPT subscription, local Ollama, OpenRouter, or LiteLLM in front of your stack — if tokens flow, traces and cost can follow.

Claude

Anthropic models via API or plans you already have.

OpenAI

GPT family and Codex-class models through the OpenAI API.

ChatGPT

Plus, Team, or Business subscription access where you use ChatGPT today.

Ollama

Run open models locally. No cloud API key required.

OpenRouter

One API key, many providers and models behind it.

LiteLLM

Proxy, routing, and fallbacks across 100+ LLM APIs in your own deployment.

› Frameworks

Multi-agent frameworks, LLM observability, and OpenClaw for gateway delivery — compose them with your models and still see traces and cost where it applies.

AutoGen

Microsoft’s multi-agent framework: agents, tools, and human-in-the-loop patterns.

CrewAI

Multi-agent crews in one dashboard.

LangChain

Chains and tools with Spotlight on top.

Langfuse

Open-source LLM engineering: traces, evals, and prompt management alongside your stack.

OpenClaw

Gateway and channels for the same agent packs — Slack, Telegram, WhatsApp, and more.

› Deploy and run

Your users are in chat apps and on a schedule — not only in the IDE. Same packs and traces when you add messaging and cron. Optional gateway-style delivery pairs with OpenClaw when you want Slack, Telegram, WhatsApp, and the rest.

Per-channel setup and the full surface list: OpenClaw channel docs →