Tinybird Agent Skills

Tinybird Agent Skills give compatible AI coding agents instructions for working with Tinybird projects. They provide context for creating and reviewing project files, SQL, tests, and deployment workflows. The agent uses the skills through the tools available in your coding environment, such as the Tinybird CLI or an SDK.

The skills follow the Agent Skills format and work with compatible coding agents such as Amp, Claude Code, Codex, Cursor, Gemini CLI, and OpenCode. Their source is available in the Tinybird Agent Skills GitHub repository.

Install Tinybird Agent Skills

Run the following command in your project:

Install Tinybird Agent Skills
npx skills add tinybirdco/tinybird-agent-skills

Update installed skills to their latest versions:

Update installed Agent Skills
npx skills update

Once installed, a compatible coding agent can select a skill when it detects a relevant task. You can also ask the agent which skills are available.

Available skills

tinybird-best-practices

Use this skill for general Tinybird project work. It covers project structure, Data Sources, Pipes, Endpoints, SQL and query optimization, deployments, testing, and secrets management.

tinybird-cli-guidelines

Use this skill when working with datafiles or running tb commands. It covers Tinybird Local, builds, deployments, data operations, Tokens, secrets, mock data, and tests.

tinybird-typescript-sdk-guidelines

Use this skill with @tinybirdco/sdk and TypeScript projects. It covers typed Data Sources and Pipes, Tinybird clients, ingestion and queries, SDK CLI workflows, and migration from datafiles.

tinybird-python-sdk-guidelines

Use this skill with tinybird-sdk and Python projects. It covers Data Sources, Pipes, Endpoints, Connections, clients, ingestion, queries, deployment commands, and migration from datafiles.

Example prompts

After installing the skills, give your coding agent a concrete project task:

  • "Create a Data Source for user events and add fixture data for local testing."
  • "Review this Pipe and optimize its Endpoint for low latency."
  • "Add tests for the Endpoints and run them with the Tinybird CLI."
  • "Define a Kafka Connection and Materialized View with the Python SDK."
  • "Validate this project and prepare it for deployment to Tinybird Cloud."

The coding agent applies the relevant skill guidance, then uses the tools and credentials available in your project to perform the task.

Agent Skills and MCP

Agent Skills provide packaged instructions and Tinybird-specific context inside a coding workflow. Use them when an agent needs to create or review project files, write SQL, add tests, or work through build and deployment commands.

The Tinybird MCP server connects an agent to a live Workspace. The configured Token limits its access, and the agent can inspect or query data and call published Endpoints. Use MCP when the task depends on live Workspace resources or data.

You can combine both. Agent Skills guide how the coding agent changes the project, while MCP lets the agent inspect the live Workspace or verify results through its permitted queries and Endpoints.

Updated
Looking for Classic docs?