AI agent resources for embedding

If you use an AI coding agent, you can give the agent Metabase-specific context to help with embedding setup, upgrades, and migrations.

Agent skills

We’ve developed some agent skills to give your AI agents step-by-step playbooks for specific embedding tasks.

Available skills

Skill	Description
SDK version upgrade	Upgrade your modular embedding SDK, including changelog checks and breaking change handling.
Full app → modular embedding	Migrate from full app embedding to modular embedding.
Modular embedding → SDK (React)	Migrate from script-based modular embedding to the React SDK.
Static → guest embeds	Migrate from static (signed) embeds to guest embeds.
SSO for embeds	Set up SSO authentication for embedded Metabase. (In development.)

Browse all skills on the agent skills repo.

llms.txt

Agents can read the docs you’re reading now (and many agents have already read our docs), but we also publish llms.txts so that you can give your AI agent:

• Summary of embedding docs, with links.
• The full embedding docs.

If you’re on a specific version (e.g., v0.58), you can use versioned llms.txt files scoped to that version’s docs:

• https://www.metabase.com/docs/v0.58/llms.txt
• https://www.metabase.com/docs/v0.58/llms-embedding-full.txt

Agents are not magic

Always review and validate the changes made by agents. Check that your application builds, tests pass, and the embedding works as expected before committing anything.

Further reading

• Embedding introduction
• Modular embedding SDK
• Upgrading the modular embedding SDK
• Agent skills repo