The Short Answer
Agentic Vibe is ibl.ai's open-source toolkit of more than 150 skills that teaches a coding agent such as Claude Code, OpenAI Codex or Cursor to build a complete enterprise AI app. Sign-in, agents, roles, memory, analytics and billing come wired in, and the app's code lands in your repository. License the platform to run it in your own infrastructure and you own all the code and the data.
What is Agentic Vibe, and what does it add to a coding agent?
Agentic Vibe is a public repository, iblai/vibe, released under the MIT license. Its latest skills release is v2.16.3, published on 26 October 2026, and the repository counts more than 150 skills grouped into ten folders.
A coding agent already knows how to write React. What it does not know is how your organization signs people in, which roles exist, where an agent's memory lives, or how usage is billed.
Each Vibe skill is a Markdown file that teaches the agent one of those things, the way a senior engineer would.
| Skill folder | What it teaches the coding agent | Skills |
|---|---|---|
| agents | Chat, browse, create and configure agents, every settings tab | 61 |
| content | Courses, catalog, credentials, admissions, avatar video | 26 |
| users | Profiles, custom user data, memories, roles and admins, invitations | 18 |
| ship | Test, deploy, hosting, native builds, app stores | 12 |
| start | The first conversation, connecting your organization, sign-in | 11 |
| organizations | Organization settings, branding, integrations | 10 |
| security | Authorized-use security reviews | 8 |
| monetization | Paywalls, item sales, subscriptions | 7 |
| billing | Credits, spend caps, pricing rails | 5 |
| analytics | Usage, topics, transcripts, costs, reports | 2 |
The counts above are the repository's own folder table at the time of writing. The repo changes often, so treat them as a snapshot, not a contract.
Which coding agents can build enterprise AI apps with Agentic Vibe?
Agentic Vibe works with the coding agent your developers already use. One install command, npx skills add iblai/vibe --all, puts the skills into Claude Code, OpenAI Codex, Cursor, OpenCode or GitHub Copilot.
Claude Code users can install it as a plugin instead, which adds the @iblai/mcp server so the agent can read the ibl.ai SDK documentation directly. Neither path asks you to switch editors or pay for a second coding agent.
How does a coding agent go from an empty folder to a working enterprise app?
The first skill, /iblai-vibe-start, asks four questions: new project or existing code, one organization, many, or none, who signs in, and what the app is about.
It then connects your ibl.ai organization through a one-click browser flow, so the API token never passes through the chat.
According to the repository, about thirty seconds after you sign in you are talking to your own agent inside an app that already has users, an admin area, custom user and organization settings, memory, analytics and billing.
From there you ask for features in plain language and the matching skill does the wiring.
Does a vibe-coded app get real SSO, RBAC and audit?
Yes, and this is the difference between a prototype and an internal tool IT will approve. Sign-in uses ibl.ai SSO, pinned to your organization.
The /iblai-vibe-rbac skill documents and mounts the platform's default roles and its Roles and Policies management screens, so access is defined by the platform, not by code the agent improvised.
Because every app sits on the same backend, every agent call lands in the same analytics, transcripts and audit trail as the rest of your ibl.ai deployment. The fiftieth app your team ships is governed exactly like the first.
Where do Agentic Vibe apps deploy, and who owns the code?
The code is yours from the first commit: it is a Next.js project in your repository, and the toolkit and scaffold are MIT-licensed, so you can fork, rename and keep them.
The /iblai-vibe-ops-deploy skill publishes it through ibl.ai's hosting, which is Vercel-backed, and returns a live URL with no Vercel account needed.
The same source builds native apps for macOS, Windows, Linux, iOS and Android through Tauri, and a release skill generates Fastlane config to submit to the App Store and Google Play.
One honest boundary: the toolkit runs against the hosted iblai.app environment by default.
To run the full ibl.ai platform behind your apps in your own cloud, on-premise or air-gapped, you license the platform codebase: the same platform that serves 1.6M+ users across 400+ organizations.
How is Agentic Vibe different from a hosted AI app builder?
Hosted builders are fast, and for a landing page or a quick prototype they are a good choice. The difference shows up when the app touches internal data and real users. The same argument runs through Code Mode, one prompt to a running Next.js app you own.
| Question | Agentic Vibe | Typical hosted builder |
|---|---|---|
| Where does the code live? | Your repository, MIT-licensed toolkit | The builder's project, usually synced to GitHub, but tied to the builder's runtime and services |
| Which coding agent? | The one you already use (Claude Code, Codex, Cursor, Copilot) | The builder's own |
| Sign-in and roles | Organization SSO and platform RBAC, wired by skills | User sign-in built in; organization SSO and audited RBAC added by you |
| Agent backend | ibl.ai agents, memory, analytics, billing | LLM calls available; agent memory, analytics and governance are yours to build |
| Platform behind the app | Hosted, or licensed to run in your own infrastructure | Vendor-hosted |
The model under the agent is configuration, not a rewrite: ibl.ai is model-agnostic, so the same app can run on Claude, GPT, Gemini, Llama or a local model.
How do you start with Agentic Vibe?
Create a free organization at ibl.ai/join, run npx skills add iblai/vibe --all in your coding agent, then type /iblai-vibe-start. The Agentic Vibe product page covers sector-specific uses, and ibl.ai's engineers can build the first app alongside your team.