Luzmo skills

Agent skills for building embedded analytics applications with Luzmo. Compatible with Cursor, Claude Code, Codex, Windsurf, Google Antigravity, GitHub Copilot, and any tool that supports the Agent Skills standard.

A set of 10 skills that turn your AI coding agent into a Luzmo embedded-analytics expert. They cover the full lifecycle: authentication and the POST + action API model, embedding saved dashboards and charts, building Flex and custom charts in code, self-service analytics (Embedded Dashboard Editor or ACK), AI / natural-language Q&A ( /aiprompt , IQ), multi-tenant isolation, data integration, resource CRUD, theming, and troubleshooting.

So your agent ships correct, secure integrations the first time instead of guessing the API and tripping over the common traps (leaked API keys, missing tenant filters, REST-verb / action: "search" mistakes).

Quick install

npx @luzmo/agent-skills@alpha

The interactive wizard installs skills into the right directories for your AI tools. Optionally register the hosted Luzmo MCP server (opt-in).

Non-interactive

npx @luzmo/agent-skills add --tool cursor --skills all --scope project --yes
npx @luzmo/agent-skills add --tool claude-code,codex --skills core,ai-analytics --yes
npx @luzmo/agent-skills add --tool cursor --with-mcp --region eu --yes
npx @luzmo/agent-skills list
npx @luzmo/agent-skills doctor

MCP server (opt-in)

Hosted MCP (EU example):

https://api.luzmo.com/0.1.0/mcp

Set credentials in your environment (never commit them):

export LUZMO_KEY="your-api-key"
export LUZMO_TOKEN="your-api-token"

US: https://api.us.luzmo.com/0.1.0/mcp — VPC: https://api.<vpc>.luzmo.com/0.1.0/mcp

Full MCP guide (tools, auth, themes, MCP App): MCP server introduction .

In practice

Ask your agent in natural language:

"Embed the logistics dashboard in this page, filtered to the current organization id."

With core + multitenancy loaded, it generates the token server-side, scopes it to the tenant, and passes only the safe values to the client:

// Server-side only (never ship API key/token to the browser)
const embed = await client.create('authorization', {
  type: 'embed',
  username: user.id,
  name: user.name,
  email: user.email,
  role: 'viewer',
  access: { /* logistics dashboard in scope */ },
  parameter_overrides: { organizationId: user.organizationId }, // row-level tenant filter
});

// Client-side: only the embed key + token
<luzmo-embed-dashboard authKey={embed.id} authToken={embed.token} />

Install options by platform

Everything ships from a single repository — there are no separate per-tool plugin repos.

Cursor (native plugin)

/add-plugin https://github.com/luzmo-official/agent-skills.git

Claude Code (native plugin)

/plugin marketplace add luzmo-official/agent-skills
/plugin install luzmo-agent-skills

Google Antigravity

npx @luzmo/agent-skills@alpha add --tool antigravity --skills all --yes
npx @luzmo/agent-skills@alpha add --tool antigravity --scope global --yes

skills.sh-compatible agents (Codex, Copilot, Gemini CLI)

npx skills add luzmo-official/agent-skills --list
npx skills add luzmo-official/agent-skills
npx skills add luzmo-official/agent-skills --skill core

Skills

Always install core at minimum.

Agents

Subagents for heavier, multi-step workflows. Delivered through the native Cursor and Claude Code plugins (auto-discovered from agents/ ).

Manual install (fallback)

git clone https://github.com/luzmo-official/agent-skills .agents/skills/agent-skills
# or symlink skills/* into .cursor/skills/, .claude/skills/, etc.

Related resources

• @luzmo/agent-skills on npm

• luzmo-official/agent-skills on GitHub

• AI Agents Documentation — llms.txt , llms-full.txt , and AGENTS.md

• MCP server introduction