omtx

Quick starts, tutorials, cookbooks, and the Om Discovery plugin for using Om from coding agents.

Use these guides to learn the Om API, work with the omtx Python SDK, run data-access workflows, launch diligence jobs, use Hub models and artifacts, and run LULA scoring with hosted Om workflows or local open-weight models.

Om MCP plugin

Add the hosted Om MCP and the Om Discovery skill to Claude Code, Codex, Cursor, Copilot, Gemini, Grok, or another compatible agent:

npx skills add omtx-ai/omtx
npx plugins add omtx-ai/omtx

Or point any remote HTTP MCP client at:

https://agents.omtx.ai/mcp

Complete the client OAuth login. Om asks for email only. Then ask for om_status. Setup details: Om MCP Overview.

Start Here

• Quick Start: first successful API, SDK, and CLI workflows.
• Tutorials: step-by-step walkthroughs for data access, diligence jobs, and Hub artifact jobs.
• Cookbooks: applied recipes for common research and modeling tasks.
• Examples: runnable Python scripts.
• Notebooks: Colab-friendly LULA workflow notebooks.

What You Can Do

• Query the Om API from curl, scripts, and notebooks.
• Install and use the omtx Python SDK.
• Load binder and non-binder datasets for accessible proteins.
• Submit diligence jobs and poll results.
• Upload artifacts, launch Hub jobs, and retrieve outputs.
• Score protein sequences and SMILES with hosted LULA-1/LULA-2 or local open-weight LULA-1, LULA-1.1, and LULA-2.
• Fetch Om Accessible Space molecules by Wallet Credit tier, score them with your own tools or LULA, and create Wallet Credits-funded Molecule Fulfillment orders.
• Use idempotency keys for safe retries.

Prerequisites

• An OMTX API key for API, hosted, and Hub workflows.
• Python 3.9+ for SDK examples.
• pip install omtx for Python workflows.
• pip install "omtx[lula]>=2.0.20" for local open-weight LULA scoring.

Keep API keys in environment variables:

export OMTX_API_KEY="your-api-key"

Choose A Path

• Agent / MCP: install the plugin above, or follow Om MCP Overview.
• New to OMTX: start with API Quickstart or Python SDK Quickstart.
• LULA scoring: follow LULA Quickstart, then use LULA Score To Order.
• Molecule selection without LULA: use Molecule Accessible Space To Order.
• Colab users: open LULA Om Accessible Space To Order or LULA Explicit SMILES.
• Loading datasets: follow Data Access Python Workflows, then use Build A Training Set.
• Running diligence: follow Diligence Jobs, then use Diligence Briefing.
• Running Hub jobs: follow Hub Jobs And Artifacts, then use Hub Artifact Workflow.