Note: These skills and servers run against the hosted api.iblai.app environment. If you'd like a license to the full platform codebase to run locally or self-host, reach out to our team at ibl.ai/contact.

Quick Start

Before you start: you need Node.js (for npx), a skills-compatible AI agent (Claude Code, Cursor, OpenCode, and 15+ others), and an ibl.ai account. No account yet? Sign up at ibl.ai/join — it walks you through creating your own organization.

1. Install the skills

npx skills add iblai/api

This installs skills that teach your AI agent how to drive the ibl.ai platform REST API directly — configure agents, manage datasets and memory, administer users and roles, send notifications, and pull analytics — for any organization you belong to. You'll now have the /iblai-* commands in your agent.

2. Connect an organization

Run the login skill once. It opens login.iblai.app/me, you sign in via SSO, it asks which of your organizations to use, and it writes your org key, username, and a Platform API Token to .env:

/iblai-login

First time? No ibl.ai account → start at ibl.ai/join (it leads you through creating your organization), then run /iblai-login. Already signed in but only see the shared main organization? That's not your own workspace — create your own org first; /iblai-login handles it.

Every other skill then reads IBLAI_ORG, IBLAI_USERNAME, and IBLAI_API_KEY from .env and calls https://api.iblai.app with Authorization: Api-Token <key>.

What is iblai/api

A toolkit for operating the ibl.ai platform from your AI agent. Where iblai/vibe gives you UI components to build an app, iblai/api gives you skills to run the platform itself — every agent-configuration and platform-admin operation mapped to its exact REST endpoints (method, URL, body) — plus a hosted MCP server for the one runtime capability that isn't a REST call: chatting with a deployed agent.

Why it matters:

• One skill per operation — each /iblai-* skill maps one capability, so changing an agent's LLM or pulling cost analytics is a / command, not a docs hunt
• Any organization — authenticate once via login.iblai.app/me, then target any of your organizations by org key + Api-Token
• Endpoint-accurate — skills carry the real api.iblai.app request shapes, so your agent calls the platform correctly the first time
• Runtime chat included — a hosted Model Context Protocol server for chatting with deployed agents (streamed responses, tool use, RAG) — no local install
• No UI required — drive the platform headless, from CI, a terminal, or any MCP-capable assistant

How It Works

1. Install — npx skills add iblai/api drops the skills into your project.
2. Connect — run /iblai-login to capture your org + username from login.iblai.app/me and store an Api-Token in .env.
3. Operate — invoke any /iblai-* skill; it fills in your org, username, and (where relevant) the agent id, then calls api.iblai.app.
4. Automate — chain skills, or connect the hosted chat MCP server to your assistant to talk to agents at runtime.

Skills

After installing, use these directly in your AI agent with / commands.

Setup

/iblai-login

Agent

/iblai-agent-create        /iblai-agent-datasets
/iblai-agent-settings      /iblai-agent-embed
/iblai-agent-sandbox       /iblai-agent-memory
/iblai-agent-access        /iblai-agent-history
/iblai-agent-llm           /iblai-agent-audit
/iblai-agent-prompts       /iblai-agent-evals
/iblai-agent-skills        /iblai-agent-chat
/iblai-agent-safety        /iblai-agent-disclaimers
/iblai-agent-privacy       /iblai-agent-tools
/iblai-agent-mcp

Organization (platform admin)

/iblai-org                 /iblai-rbac
/iblai-management       /iblai-crm
/iblai-integrations     /iblai-notifications
/iblai-tokens          /iblai-invites

Profile

/iblai-profile             /iblai-profile-metadata

Content & discovery

/iblai-search              /iblai-course-create
/iblai-analytics

What each skill does

Skills live in skills/. Read them, extend them, or write your own.

MCP Server

Hosted Model Context Protocol server — no local installation required. This covers the one runtime capability the skills can't: actually talking to a deployed agent. Wire it up with /iblai-agent-chat (it writes the config below from your .env token + a chosen agent), or add it manually. (Administering the platform is the skills' job — see Skills vs MCP servers.)

Connect from Claude Code

claude mcp add iblai-agent-chat --transport http https://asgi.data.iblai.app/mcp/agent-chat/ --header "Authorization: Api-Token YOUR_API_TOKEN"

Connect from Claude Desktop / Cursor

{
  "mcpServers": {
    "iblai-agent-chat": {
      "transport": "streamable-http",
      "url": "https://asgi.data.iblai.app/mcp/agent-chat/",
      "headers": {
        "Authorization": "Api-Token YOUR_API_TOKEN"
      }
    }
  }
}

Skills vs MCP servers

The split is administer/use-via-REST vs chat at runtime:

• Skills do everything you can reach over REST — create and configure agents, manage datasets, memory, users, roles, notifications, discovery/search, recommendations, user profiles and analytics, and reporting — by calling the API directly.
• The MCP server is for the one thing that isn't a REST admin call: holding a live conversation with a deployed agent (streamed responses, tool use, RAG).

Rule: if a skill covers it, there is no server for it. That's why only iblai-agent-chat remains — the former analytics, agent-create, search, and user servers were removed once skills covered analytics, agent creation, discovery + recommendations (/iblai-search), and user profile + analytics (/iblai-profile).

Authentication

Everything authenticates the same way:

• Base URL: https://api.iblai.app
• Header: Authorization: Api-Token <key> on every request
• Org & username: from login.iblai.app/me — each organization you belong to shows its key (e.g. enterprise, iblai, or a UUID)
• Api-Token: your first token comes from the platform admin (or login.iblai.app); once you're connected, /iblai-tokens lists, creates, and rotates tokens. The secret is shown once.

The /iblai-login skill walks through all of this and writes IBLAI_ORG, IBLAI_USERNAME, and IBLAI_API_KEY to .env. Never commit .env — it is in .gitignore.

Resources

• skills.sh/iblai/api — install skills with npx skills add iblai/api
• iblai/vibe — companion toolkit for building ibl.ai apps (UI components + Claude Code skills)
• login.iblai.app/me — your account, organizations, and org keys
• docs.ibl.ai — platform documentation

License

MIT — see LICENSE. © ibl.ai