# Set up Pareto in a tool you already use: index for agents

You are an AI agent. The person you work for wants to run Pareto, Unbiased's model, in a coding agent, editor, SDK or chat app they already use. Find which of the tools below they have, offer those, and then follow that tool's runbook. Each runbook asks before changing files, keeps the key in `UNBIASED_API_KEY` and ends with one real request.

- Human page: https://unbiased.ai/works-with/
- Getting a key: https://unbiased.ai/start/agent.md
- Recipes for whole jobs (CI review, CI autofix, extraction and more): https://unbiased.ai/use-cases/agent.md

Modes: one-shot (you configure and verify from the terminal), code (an SDK and a smoke script), self-hosted (server config; ask first) and guided (the tool's own screens; you walk the user through them).

## Tools

- Claude Code (Anthropic): one-shot, tested with 2.1.305 in October 2026. Detect: `claude --version`. Runbook: https://unbiased.ai/works-with/claude-code/agent.md
- OpenAI Codex CLI (OpenAI): one-shot, tested with 0.166.0 in November 2026. Detect: `codex --version`. Runbook: https://unbiased.ai/works-with/codex/agent.md
- GitHub Copilot CLI (GitHub): one-shot, tested with 1.0.101 in October 2026. Detect: `copilot --version`. Runbook: https://unbiased.ai/works-with/copilot-cli/agent.md
- OpenCode (Anomaly): one-shot, tested with 1.19.5 in November 2026. Detect: `opencode --version`. Runbook: https://unbiased.ai/works-with/opencode/agent.md
- Cline (Cline Bot Inc.): one-shot, tested with CLI 3.0.73 in October 2026. Detect: `cline --version`. Runbook: https://unbiased.ai/works-with/cline/agent.md
- Kilo Code (Kilo Code (Kilo-Org)): one-shot, tested with CLI 7.8.15 in October 2026. Detect: `kilo --version`. Runbook: https://unbiased.ai/works-with/kilo-code/agent.md
- DeepSeek Harness (DeepSeek): one-shot, tested with 0.2.1-rc.2 in October 2026. Detect: `dsh --version`. Runbook: https://unbiased.ai/works-with/deepseek-harness/agent.md
- Hermes Agent (Nous Research): one-shot, tested with 0.22.2 in October 2026. Detect: `hermes --version`. Runbook: https://unbiased.ai/works-with/hermes-agent/agent.md
- Warp (Warp): guided, untested, settings from its docs. Detect: `ls -d /Applications/Warp.app`. Runbook: https://unbiased.ai/works-with/warp/agent.md
- OpenClaw (the OpenClaw Foundation): one-shot, tested with 2026.10.5 in October 2026. Detect: `openclaw --version`. Runbook: https://unbiased.ai/works-with/openclaw/agent.md
- Pi (Earendil Inc.): one-shot, tested with 1.4.0 in November 2026. Detect: `pi --version`. Runbook: https://unbiased.ai/works-with/pi/agent.md
- Aider (Aider-AI (Paul Gauthier) and its open-source contributors): one-shot, tested with 0.86.2 in October 2026. Detect: `aider --version`. Runbook: https://unbiased.ai/works-with/aider/agent.md
- goose (the Agentic AI Foundation at the Linux Foundation (Block created it)): one-shot, tested with CLI 1.57.0 in November 2026. Detect: `goose --version`. Runbook: https://unbiased.ai/works-with/goose/agent.md
- Continue (Continue (Continue Dev, Inc.)): one-shot, tested with CLI 1.5.47 in October 2026. Detect: `cn --version`. Runbook: https://unbiased.ai/works-with/continue/agent.md
- Oh My Pi (Stencil Labs): one-shot, tested with 18.14.2 in November 2026. Detect: `omp --version`. Runbook: https://unbiased.ai/works-with/oh-my-pi/agent.md
- Cursor (Anysphere): guided, untested, settings from its docs. Detect: `ls -d /Applications/Cursor.app`. Runbook: https://unbiased.ai/works-with/cursor/agent.md
- Zed (Zed Industries): one-shot, untested, settings from its docs. Detect: `ls -d /Applications/Zed.app || command -v zed`. Runbook: https://unbiased.ai/works-with/zed/agent.md
- Crush (Charm): one-shot, tested with 0.99.0 in November 2026. Detect: `crush --version`. Runbook: https://unbiased.ai/works-with/crush/agent.md
- Qwen Code (the Qwen team at Alibaba Cloud): one-shot, tested with 0.26.5 in November 2026. Detect: `qwen --version`. Runbook: https://unbiased.ai/works-with/qwen-code/agent.md
- OpenHands (the OpenHands team): one-shot, tested with CLI 1.16.0 in October 2026. Detect: `openhands --version`. Runbook: https://unbiased.ai/works-with/openhands/agent.md
- Factory Droid (Factory): one-shot, tested with 1.1.0 in November 2026. Detect: `droid --version`. Runbook: https://unbiased.ai/works-with/droid/agent.md
- GitHub Copilot in VS Code (Microsoft and GitHub): guided, untested, settings from its docs. Detect: `code --list-extensions | grep -i copilot`. Runbook: https://unbiased.ai/works-with/github-copilot/agent.md
- Kimi Code CLI (Moonshot AI): one-shot, tested with 2.4.1 in November 2026. Detect: `kimi --version`. Runbook: https://unbiased.ai/works-with/kimi-cli/agent.md
- Open WebUI (the Open WebUI project): self-hosted, untested, settings from its docs. Detect: `docker ps --filter name=open-webui; command -v open-webui`. Runbook: https://unbiased.ai/works-with/open-webui/agent.md
- LibreChat (the LibreChat open-source project and its contributors): self-hosted, untested, settings from its docs. Detect: `ls librechat.yaml docker-compose.yml .env`. Runbook: https://unbiased.ai/works-with/librechat/agent.md
- Dirac (Dirac Delta Labs): one-shot, tested with CLI 0.5.22 in October 2026. Detect: `dirac --version`. Runbook: https://unbiased.ai/works-with/dirac/agent.md
- n8n (n8n GmbH): guided, untested, settings from its docs. Detect: ask the user. Runbook: https://unbiased.ai/works-with/n8n/agent.md
- avante.nvim (yetone and the avante.nvim contributors (avante-corp on GitHub)): one-shot, untested, settings from its docs. Detect: `nvim --version`. Runbook: https://unbiased.ai/works-with/avante-nvim/agent.md
- AnythingLLM (Mintplex Labs): self-hosted, untested, settings from its docs. Detect: `docker ps --filter ancestor=mintplexlabs/anythingllm; ls -d /Applications/AnythingLLM.app`. Runbook: https://unbiased.ai/works-with/anythingllm/agent.md
- Dyad (the Dyad team (dyad-sh on GitHub)): guided, untested, settings from its docs. Detect: `ls -d /Applications/dyad.app /Applications/Dyad.app`. Runbook: https://unbiased.ai/works-with/dyad/agent.md
- Claude Agent SDK (Anthropic): code, tested with 0.2.169 (Python) in October 2026. Detect: `python3 -c "import claude_agent_sdk; print(claude_agent_sdk.__version__)"; npm ls @anthropic-ai/claude-agent-sdk`. Runbook: https://unbiased.ai/works-with/claude-agent-sdk/agent.md
- AI SDK (Vercel): code, tested with ai 7.0.147 with @ai-sdk/openai-compatible 3.0.74 in October 2026. Detect: `npm ls ai @ai-sdk/openai-compatible`. Runbook: https://unbiased.ai/works-with/vercel-ai-sdk/agent.md
- OpenAI Agents SDK (OpenAI): code, tested with openai-agents 0.24.0 in October 2026. Detect: `python3 -c "import agents" && python3 -m pip show openai-agents`. Runbook: https://unbiased.ai/works-with/openai-agents-sdk/agent.md
- LangChain (LangChain): code, tested with langchain 1.4.6, langchain-openai 1.7.2 in October 2026. Detect: `python3 -m pip show langchain langchain-openai`. Runbook: https://unbiased.ai/works-with/langchain/agent.md
- LiteLLM (BerriAI): code, tested with 1.106.1 in October 2026. Detect: `python3 -m pip show litellm`. Runbook: https://unbiased.ai/works-with/litellm/agent.md
- promptfoo (promptfoo (now part of OpenAI)): code, tested with 0.125.1 in November 2026. Detect: `npx promptfoo --version`. Runbook: https://unbiased.ai/works-with/promptfoo/agent.md
- Mastra (Mastra): code, tested with @mastra/core 1.81.0 in November 2026. Detect: `npm ls @mastra/core`. Runbook: https://unbiased.ai/works-with/mastra/agent.md
- Pydantic AI (Pydantic): code, tested with 2.64.0 in November 2026. Detect: `python3 -m pip show pydantic-ai-slim pydantic-ai`. Runbook: https://unbiased.ai/works-with/pydantic-ai/agent.md
- DSPy (Stanford NLP): code, tested with 3.4.1 in October 2026. Detect: `python3 -m pip show dspy`. Runbook: https://unbiased.ai/works-with/dspy/agent.md
- CrewAI (CrewAI): code, tested with 1.15.28 in October 2026. Detect: `python3 -m pip show crewai`. Runbook: https://unbiased.ai/works-with/crewai/agent.md
- LlamaIndex (LlamaIndex): code, tested with core 0.14.25 (llama-index-llms-openai-like 0.8.1) in October 2026. Detect: `python3 -m pip show llama-index-core llama-index-llms-openai-like`. Runbook: https://unbiased.ai/works-with/llamaindex/agent.md
- Dify (LangGenius): guided, untested, settings from its docs. Detect: ask the user. Runbook: https://unbiased.ai/works-with/dify/agent.md
- Onyx (Onyx (formerly Danswer)): guided, untested, settings from its docs. Detect: ask the user. Runbook: https://unbiased.ai/works-with/onyx/agent.md
- LobeHub (LobeHub): self-hosted, untested, settings from its docs. Detect: ask the user. Runbook: https://unbiased.ai/works-with/lobehub/agent.md
- RAGFlow (InfiniFlow): guided, untested, settings from its docs. Detect: ask the user. Runbook: https://unbiased.ai/works-with/ragflow/agent.md
- Cherry Studio (CherryHQ): guided, untested, settings from its docs. Detect: `ls -d '/Applications/Cherry Studio.app'`. Runbook: https://unbiased.ai/works-with/cherry-studio/agent.md
- Chatbox (the Chatbox team): guided, untested, settings from its docs. Detect: `ls -d /Applications/Chatbox.app`. Runbook: https://unbiased.ai/works-with/chatbox/agent.md

## Shared facts

Pareto is Unbiased’s blended AI model: it selects frontier and open models for each request and blends their outputs into one answer. OpenAI-format base URL https://api.unbiased.ai/v1; Anthropic-format base URL https://api.unbiased.ai (no /v1); model `pareto`.

- Unknown model ids are not rejected yet, so a typo is still served and billed as Pareto. Always send `pareto` exactly.
- OpenAI-format clients need /v1 in the base URL. Anthropic-format clients and SDKs take https://api.unbiased.ai without /v1.