Dedalus Skills

Skills for building AI agents and MCP servers with the Dedalus platform. Works with any coding agent or IDE that supports skills.

Skills Included

Skill	Description
use-dedalus-sdk	Build AI-powered applications with dedalus_labs — agent orchestration, tools, structured outputs, streaming, multi-provider models
use-dedalus-mcp	Build MCP servers with dedalus_mcp — tools, resources, prompts, context utilities

Installation

Install with the skills CLI:

npx skills add adamsardo/dedalus-skills

This installs all skills in the repo (currently use-dedalus-mcp and use-dedalus-sdk). The name value in each SKILL.md becomes the slash command in agents that support skills (e.g., /use-dedalus-mcp).

Install a single skill if you only want one:

npx skills add adamsardo/dedalus-skills --skill use-dedalus-mcp

What's Inside

use-dedalus-sdk

For building AI agents with the Dedalus SDK:

from dedalus_labs import AsyncDedalus, DedalusRunner

client = AsyncDedalus()
runner = DedalusRunner(client)

result = await runner.run(
    input="Search for AI news and summarize",
    model="anthropic/claude-sonnet-4-20250514",
    mcp_servers=["windsor/brave-search-mcp"],
)

Covers:

• DedalusRunner — Agent orchestration with automatic tool execution
• Tools — Local functions with auto-generated schemas
• Structured Outputs — Type-safe JSON with Pydantic/Zod
• Streaming — Real-time responses with stream=True
• MCP Integration — Connect to any MCP server
• Handoffs — Multi-model routing
• Multi-provider support — OpenAI, Anthropic, Google, xAI, DeepSeek, Mistral

use-dedalus-mcp

For building MCP servers:

from dedalus_mcp import MCPServer, tool

@tool(description="Add two numbers")
def add(a: int, b: int) -> int:
    return a + b

server = MCPServer("calculator")
server.collect(add)

asyncio.run(server.serve())

Covers:

• @tool — Functions the LLM calls during reasoning
• @resource — Data the LLM can read for context
• @prompt — User-selectable message templates
• get_context() — Logging, progress, cancellation
• MCPServer — Modular server setup with collect()
• Testing patterns

Usage Examples

Once installed, Claude will automatically use these skills when you ask about Dedalus:

"Help me build an AI agent that searches the web and summarizes results"

"Create an MCP server with tools for database queries"

"How do I use structured outputs with Pydantic in Dedalus?"

"Show me how to stream responses from DedalusRunner"

Requirements

• Python 3.10+ or Node.js 18+
• Dedalus API key (DEDALUS_API_KEY)

# Install the SDKs
pip install dedalus-labs dedalus-mcp  # Python
npm install dedalus-labs              # TypeScript

Documentation

• Dedalus SDK Docs
• Dedalus MCP Docs
• API Reference

Contributing

Found an issue or want to improve these skills? PRs welcome!

1. Fork the repo
2. Edit the skill files in use-dedalus-sdk/ or use-dedalus-mcp/
3. Test with Claude
4. Submit a PR

License

MIT — see LICENSE for details.

---

Built with ❤️ for the Dedalus community