Welcome to Skilder
Build, share, and run capabilities for AI agents. Bundle MCP tools with instructions into reusable skills your agents discover via MCP.
Skilder helps you build capabilities for your AI agents. You bundle MCP tools with instructions, scripts, and reference docs into skills — structured capabilities that agents discover and use. Your agents connect to Skilder via MCP, and Skilder handles the rest: running the tools, checking permissions, and routing requests.
When you create a workspace, the Skilder Assistant is set up automatically — a built-in AI that helps you discover automation opportunities and co-create skills through conversation. You bring your own AI keys, co-create skills with the assistant, and connect your agent via MCP. That's it.
We're in public beta. Things move fast, and we're building this with you. If something breaks or feels off, let us know.
Get Started in 5 Minutes
Your workspace comes pre-configured. Co-create a skill and connect your agent.
Connect Your Agent
Hook up Claude, ChatGPT, Copilot, or any MCP client.
Write a Script
Call tools, spawn sub-agents, and chain scripts from TypeScript or Python.
How It Works
Sign up → Skilder Assistant configures your workspace → Co-create skills → Connect agent via MCP- Sign up — Create a workspace. The Skilder Assistant is configured automatically with built-in skills to help you get started.
- Co-create skills — Chat with the Skilder Assistant in the Studio to discover use cases and build skills together. Or create them manually with full control.
- Install tools — Add MCP servers from the registry or paste your own config. Skilder runs them for you.
- Connect via MCP — Your agent connects to Skilder as an MCP server (via STDIO or Streamable HTTP) and discovers all available capabilities automatically.
What Makes Skilder Different
Capabilities, Not Just Tools
Go beyond raw tool access. Skills bundle tools with instructions, scripts, and assets — so agents know what to do, not just what they can call.
Your Agent Connects via MCP
Skilder is an MCP server. Claude, ChatGPT, Copilot, or any MCP client connects directly — STDIO and Streamable HTTP supported.
Bring Your Own Keys
Use your existing API keys for Anthropic, OpenAI, Google AI, Infomaniak, or Ollama. No vendor lock-in, no markup.
Cloud Runtime
MCP servers run in Skilder's managed infrastructure. No Docker, no local setup — capabilities are always available.
Script SDK
Write TypeScript or Python scripts inside skills that call other tools, spawn sub-agents, and chain workflows together.
Team-Ready
Workspaces, role-based access (Admin, Editor, Viewer), and shared configurations for your whole team.
Next Steps
- Quickstart — Co-create your first skill and connect your agent in 5 minutes.
- Connect Your Agent — Link Claude, ChatGPT, Copilot, or any MCP client via MCP.
- AI Providers — Set up your Anthropic, OpenAI, or Google AI keys.
- Skills — Understand skills, hats, and execution modes.
- Tools — Browse, install, and configure MCP servers.
- Studio — Test your skills interactively.
- Writing Scripts — Use the Script SDK to build powerful workflows.
- Monitoring — Track tool calls and agent performance.