DEVELOPERS API, MCP and agent-ready workflows.
humiris
Book demoGet started
BUILD WITH HUMIRIS

Connect Humiris to
the way you build.

Use one familiar API in your product, or give Claude Code, Cursor, Codex and other MCP clients the context to implement it with you.

Create an API key Platform quick start
PICK AN INTEGRATION

Copy less.
Ship with context.

Switch between application code, the live Humiris MCP, a project skill and a complete coding-agent prompt. Every example is ready to copy.

OpenAI-compatible APIKeep your SDK. Change the key, base URL and model route.
import OpenAI from "openai";

const humiris = new OpenAI({
  apiKey: process.env.HUMIRIS_API_KEY,
  baseURL: "https://api.humiris.ai/api/v1",
});

const response = await humiris.chat.completions.create({
  model: "humiris/auto",
  messages: [
    { role: "user", content: "Summarize this support case" },
  ],
});

console.log(response.choices[0].message.content);
CODING AGENTS

Meet developers
where they work.

Choose your coding agent or IDE. Every client gets a focused setup, real logo, official documentation and a configuration you can copy.

CLI

Claude Code

Add the Humiris remote MCP at project scope in one terminal command.

Docs
MCP

Codex

Expose Humiris model search, routing and integration tools inside Codex.

Docs
MCP

Cursor

Give Cursor Agent access to Humiris from a project MCP configuration.

Docs
MCP

Windsurf

Connect Cascade to the remote Humiris MCP from Windsurf settings.

Docs
MCP

JetBrains

Use Humiris from AI Assistant in IntelliJ, PyCharm, WebStorm and more.

Docs
MCP

Zed

Bring Humiris tools into Zed's agent panel through a context server.

Docs
MCP

GitHub Copilot

Use Humiris with Copilot agent mode and install project-scoped skills.

Docs
MCP

Gemini CLI

Connect Google's terminal agent to the Humiris remote tool catalog.

Docs
CONNECT

Claude Code

Run claude mcp list after installation to verify the server.

claude mcp add --transport http humiris --scope project https://humiris.ai/api/mcp
AGENT SKILLS

Reusable knowledge.
Plain Markdown.

Install focused Humiris skills using the open SKILL.md format. Your agent discovers the right workflow only when the task needs it.

CHOOSE YOUR AGENT
INSTALL · HUMIRIS
mkdir -p .agents/skills/humiris
curl -L https://humiris.ai/developers/skills/core.md \
  -o .agents/skills/humiris/SKILL.md
1

InstallCopy the Markdown skill into your project.

2

Auto-discoverYour agent reads the SKILL.md when relevant.

3

Use itAsk the agent to build with Humiris.

Download SKILL.md
READY PROMPT

Give the task to your coding agent.

This brief tells the agent what to inspect, how to protect your key, where to isolate the integration and what evidence to return.

Download Humiris skill
Integrate Humiris into this repository as a production-ready AI gateway.

First inspect the existing architecture, package manager, API layer and test setup. Then:
1. Use the existing OpenAI SDK when available; otherwise add its official SDK.
2. Read HUMIRIS_API_KEY only from the server environment. Never expose or log it.
3. Set the base URL to https://api.humiris.ai/api/v1.
4. Use humiris/auto as the default route, while making the model configurable.
5. Add one small adapter/service instead of coupling Humiris throughout the app.
6. Preserve current behavior, types, lint rules and UI conventions.
7. Add useful timeout/error handling and one real smoke or integration test.
8. Run the relevant checks. Report changed files, results and exact local setup steps.

Do not claim success unless the code builds and the test passes.
Keep keys server-side. Store HUMIRIS_API_KEY in your secret manager or environment. Never place it in browser code, commits or copied prompts.