AI Ecosystem
CLI, knowledge files, skills, and MCP servers powering the most AI-integrated app platform
Most platforms treat AI as an add-on. Runwork is built around it. Every app auto-generates SKILL.md for any AI tool. During local development, the platform creates CLAUDE.md + AGENTS.md on the fly so your AI editor understands the framework. Every app automatically becomes an MCP server. Skills turn app capabilities into AI expertise. Connect Claude Code, Cursor, Codex, or any MCP-compatible tool and they get full read and write access. One ecosystem where everything connects.
Powered By
The Challenge
- AI code assistants generate generic code that doesn't understand your platform or framework
- AI tools are siloed: your chatbot, code assistant, and automation tool don't talk to each other
- Business knowledge lives in people's heads, not in the AI's context
- Connecting AI tools to your actual business data requires custom integrations
- Every new AI tool means another setup, another API key, another context to manage
- No single platform covers the full spectrum from quick answers to custom apps
The Runwork Solution
- CLAUDE.md + AGENTS.md are created on the fly during development, SKILL.md is auto-generated for every app
- Download SKILL.md from the dashboard or develop locally where the platform provides all knowledge files
- Every app automatically becomes an MCP server with full read and write access
- Skills layer your business knowledge on top: best practices, recipes, guardrails
- MCP Client brings external AI capabilities into your workspace
- One ecosystem where apps, skills, CLI, and connections compound over time
What You Can Build
Claude Code + Full Framework Knowledge
Run `runwork clone my-crm`, open in Claude Code. The platform creates CLAUDE.md + AGENTS.md on the fly during `runwork dev`, and SKILL.md documents your app's entities, endpoints, and workflows. AI-generated code is framework-aware from the first prompt.
Every App is an MCP Server
Your CRM app automatically exposes MCP tools for managing contacts, triggering workflows, uploading files. Connect Claude Desktop, Cursor, or any MCP client and interact with your app's full capabilities.
Portable Skills
Download any app's SKILL.md from the dashboard and use it in Cursor while coding. Your IDE understands your business data model, endpoints, and workflows.
CRM That Teaches AI Sales
Your CRM becomes a skill. The AI knows your pipeline, your process, your close rates. Load it in #sales with a slash command, or download SKILL.md for your local editor.
Cross-App Intelligence
Load your CRM skill and Marketing skill in the same channel. The AI orchestrates across both: "Draft a follow-up for leads who attended the webinar."
Local Dev to Production
`runwork init` creates your app. `runwork dev` boots a sandbox and the platform provides CLAUDE.md + AGENTS.md live. `runwork deploy` to ship. Your team uses it immediately.
External Knowledge In
Connect a company knowledge base MCP server. Your workspace AI can now search internal docs alongside your app data in the same conversation.
Channel Expertise
Pin the Helpdesk and CRM skills to #customer-success. The AI always has the right context. No slash commands needed, just ask.
A Day in the Life
8:30 AM - Build locally with AI. You run runwork clone inventory-tracker and open the project in Claude Code. Run runwork dev and the platform creates CLAUDE.md + AGENTS.md on the fly, giving Claude full context about the Runwork framework. SKILL.md documents your app's entities, endpoints, workflows, and integrations. You tell Claude: "Add a low-stock alert that triggers a Slack notification." Claude generates correct framework code on the first try because it understands every API.
9:30 AM - Live preview, zero config. runwork dev boots a remote sandbox and streams your changes. You see the low-stock alert working in your browser while Claude Code keeps iterating. No Docker, no local database, no runtime to manage. The CLI handles everything.
11:00 AM - The team uses it. runwork deploy. Done. Your team opens the dashboard, and the Inventory Tracker is updated. In #operations, a teammate asks the AI: "Which products are running low?" The AI loads the Inventory Tracker skill and queries real data. The same app you just coded locally is now team expertise.
1:00 PM - External AI tools connect. A colleague opens Claude Desktop, connected to the workspace via MCP. They ask: "List all products below reorder threshold and draft purchase orders." Claude calls the Inventory Tracker's MCP tools, queries the data, and generates orders. Every app is automatically an MCP server with full read and write access.
2:30 PM - Cross-app intelligence. In #sales, someone loads the CRM and Inventory skills together. "Which customers ordered products that are running low? Draft a heads-up email." The AI orchestrates across both apps, correlates data, and generates personalized emails. Two apps, one conversation.
4:00 PM - The flywheel. You created one app. It automatically became: a skill available in any channel, an MCP server accessible to Claude Desktop and Cursor, a SKILL.md downloadable for any AI tool, and a team tool on the dashboard. One app, the entire ecosystem gets smarter. That's the Runwork AI ecosystem.
Frequently Asked Questions
What makes Runwork the "most AI-integrated app platform"?
What are CLAUDE.md, AGENTS.md, and SKILL.md?
runwork dev, providing project context and framework instructions to AI tools like Claude Code and Codex. Together they give any AI tool complete understanding of both the framework and your app.How do CLI, skills, and MCP connect?
Which AI tools work with Runwork?
Do I need to set up skills and MCP separately?
Related Use Cases
Ready to build?
Start a 14-day free trial. No credit card required.