Quickstart
Get up and running with Skilder in 5 minutes. Your workspace comes pre-configured — co-create your first skill and connect your agent.
When you create a workspace, Skilder automatically sets up a Skilder Assistant — a pre-configured hat with built-in skills that help you discover automation opportunities and build new capabilities. This guide walks you from sign-up to a working skill connected to your favorite AI client.
What you will do:
- Sign up and let the assistant configure your workspace
- Connect an AI provider so the Studio can power conversations
- Co-create a skill with the Skilder Assistant
- Connect your agent via MCP and test it from a third-party client
Time required: About 5 minutes.
Step 1: Sign Up and Get Your Workspace
Go to app.skilder.ai and create an account with your email address. After signing in, create your first Workspace.
Once your workspace is ready, Skilder's onboarding bot automatically configures the Skilder Assistant — a hat pre-loaded with skills for discovering automation opportunities, analyzing workflows, and guiding you through skill creation. You don't need to build anything from scratch to get started.
If you were invited to an existing workspace, accept the invitation instead — the assistant is already set up.
Step 2: Connect Your AI
Link an AI provider so Skilder can power the Studio and help you build skills interactively. Skilder uses a BYOK (Bring Your Own Key) model — you connect your own API keys.
- From the dashboard, click Connect Provider.
- Choose your provider — Anthropic, OpenAI, Google AI, Infomaniak, or Ollama.
- Click Configure on the provider card.
- Paste your API key and any required settings.
- Save the configuration.
Need help getting an API key? See the full AI Providers guide.
Step 3: Co-Create Your First Skill
Now that your workspace is ready and your AI provider is connected, open the Studio from the top navigation bar and select the Skilder Assistant hat.
The Skilder Assistant helps you build skills through conversation. Describe what you need, and it will guide you through the process. For example:
I need a skill that summarizes our weekly Slack threads
and posts a digest to our #team-updates channel.The assistant will:
- Clarify the use case — ask about your workflow, who benefits, and what the output should look like.
- Suggest a skill structure — propose a name, description, instructions, and which tools are needed.
- Create the skill — set up the skill with the right tools and instructions, ready for you to review and refine.
Once the assistant has drafted the skill, open it in the Skill Editor to fine-tune the instructions, adjust tools, or change the execution mode. See Skill Editor for details.
Tip: You can also create skills manually from the Skills page or from within a hat. See Create Your First Skill for a quick reference or the full Manually Compose a Skill tutorial for a hands-on walkthrough.
Step 4: Install Tools (If Needed)
The Skilder Assistant may suggest tools that are not yet installed in your workspace. If so, add them from the tool registry:
- Click Tools in the top navigation bar, then click Add Tools.
- Find the suggested MCP server in the registry (e.g.,
@anthropic-ai/server-slack). - Click the card, review the configuration, click Test Server to verify it works, then click Finish.
- Wait for the server to show a Connected status.
Once the tools are installed, link them to your skill from the skill editor's settings panel.
Many tools work out of the box with no API keys. Others — like Slack, GitHub, or Google Workspace — require OAuth or environment variables. See Tools for details.
Step 5: Connect Your Agent via MCP
Skilder acts as an MCP server. Your agent connects to it the same way it connects to any MCP server.
Click the Connect button from the top navigation bar or from a hat card. The dialog generates the right configuration for your client.
| Client | Transport | How to Connect |
|---|---|---|
| Claude | STDIO | Click Download .mcpb — open the file and Claude Desktop configures itself automatically. |
| Microsoft Copilot | Streamable HTTP | Copy the MCP endpoint URL into Copilot Studio. |
| OpenAI ChatGPT | Streamable HTTP | Copy the MCP endpoint URL into ChatGPT settings. |
| Other Clients | STDIO or Streamable HTTP | Any MCP-compatible client — use the STDIO config JSON or endpoint URL shown in the dialog. |
If you don't have an API key yet, create one directly from the dialog. See Connect Your Agent for full details and API Keys for key management.
Step 6: Test From Your Agent
Open your connected AI client and try out the skill you just built. For example:
Summarize the latest threads in #team-updates.The agent should discover the skill, call the right tools, and return results. If it works — you've got a live, reusable capability that any team member can connect to.
Verify the connection:
What capabilities do you have available?The agent should list the skills and tools available through its hats.
You're Live
Your workspace is set up, your first skill is running, and your agent is connected. From here, the loop is simple: identify a use case → co-create a skill → connect and test.
What to Do Next
- Create Your First Skill — Build skills manually with full control over instructions, tools, and scripts.
- Studio — Test and refine skills interactively.
- Skills & Hats — Organize capabilities into roles like "Sales Rep" or "Data Analyst."
- Tools — Install more MCP servers from the registry.
- Writing Scripts — Add TypeScript or Python scripts for advanced workflows.
- OAuth Integrations — Connect OAuth-based services like Slack, GitHub, or Google Workspace.
- Monitoring — Track tool calls, agent connections, and performance.
- Concepts — Understand Skilder's architecture and data flow.
- Troubleshooting — Fix common issues.