Official Cube skills for Claude Code, Cursor, OpenAI Codex, GitHub Copilot, Gemini CLI, and other Agent Skills compatible agents.
Explore and build your semantic model, manage workbooks and dashboards, run queries, administer access, and ship deployments — from the agent you already work in. The skills drive the Cube CLI, so anything you can do through Cube's public API, your agent can do too.
Cube agent skills run in your coding agent and operate Cube through the CLI. They are not Agent Skills in Cube, which your data team authors in the semantic model and runs from Analytics Chat.
Claude Code
/plugin marketplace add cube-js/cube-agent-skills
/plugin install cube@cube
Codex, Copilot, Gemini CLI, and other skills.sh-compatible agents
npx skills add cube-js/cube-agent-skills
Cursor and Snowflake Cortex Code are coming next; until then, copy skills/
into that agent's skills directory.
Install the Cube CLI:
curl -fsSL https://raw.githubusercontent.com/cube-js/cube/master/install-cli.sh | shThen authenticate, either way:
cube login # interactive — opens a browser
export CUBE_API_URL=... CUBE_API_KEY=... # headless, CI, agent loopsEvery skill checks both before doing anything, and stops with instructions rather than guessing.
| Skill | What it does |
|---|---|
cube-explore-model |
Search and inspect the semantic model — cubes, views, measures, joins, and impact analysis before a change |
cube-build-model |
Author cubes and views in YAML on a dev-mode branch, validate, commit, deploy |
cube-explore-content |
Browse workbooks, dashboards, reports and folders |
cube-build-content |
Create and update workbooks, reports, dashboards and scheduled notifications |
cube-run-query |
Run semantic-layer queries and interpret the results |
cube-configure-agent |
Inspect and tune the in-product agent — agents, rules, certified queries, skills |
cube-admin |
Users, groups, attributes, access policies, tenant settings, SCIM and OIDC |
cube-embed |
Embed sessions, tokens and embed tenants for embedded analytics |
cube-deploy |
Deployments, environments, environment variables, build status and logs |
Skills activate on their own when a request matches. You can also name one directly: "use cube-build-model to add a churn measure."
Cube has three agent-facing surfaces and the names are close enough to be worth stating plainly:
| What it is | Where it runs | |
|---|---|---|
| Cube connector | Ask Cube questions in natural language | Claude — desktop, web, Code |
| Cube agent skills (this repo) | Operate Cube — model, content, access, deployments | Your coding agent, over the cube CLI |
| Agent Skills in Cube | Saved workflows your data team authors in the semantic model | Analytics Chat, via the / menu |
The Cube connector and these skills compose: connect both, and Claude can ask questions of your data and act on the platform.
Issues and pull requests welcome — see CONTRIBUTING.md.
Commits need a DCO sign-off (git commit -s), the same as
cube-js/cube.