A community-driven registry for Claude, Cursor, Windsurf, Cline & more. Not affiliated with Anthropic.
Are you the author? Sign in to claim
Engineering reliable analytics agents from scratch with DuckDB, MCP and a semantic layer
PyCon DE & PyData 2026 — April 14, 11:45, Dynamicum [Ground Floor] 90-minute hands-on tutorial by Mehdi Ouazza and Dumky de Wilde
workshop/ <-- START HERE
├── duckoffee.duckdb # Dataset: a coffee-shop chain across 13 cities
├── skill.md # Agent instructions (semantic layer)
├── pyproject.toml # Python deps: duckdb, pydantic-ai, fastmcp, openai
├── Makefile # make setup
└── (your code) # agent_loop.py, agent_tools.py, mcp-server/, …
solutions/ # Reference implementations for every part
site/ # Slide deck (Next.js → GitHub Pages)
└── src/content/*.mdx # One MDX file per part
.devcontainer/ # Codespace / devcontainer (Python, uv, opencode)
All exercises happen inside workshop/. Clone the repo, cd workshop, and follow the slides.
git clone https://github.com/motherduckdb/analytics-agent-duckdb-workshop.git
cd analytics-agent-duckdb-workshop/workshop
You need Python 3.10+, uv, and opencode:
# uv (fast Python package manager)
curl -LsSf https://astral.sh/uv/install.sh | sh
# opencode (agentic CLI)
curl -fsSL https://opencode.ai/install | bash
# workshop Python deps
uv sync
Pick one:
| Option | How |
|---|---|
| MotherDuck-minted key (recommended) | Go to motherduck.com/minter — pre-funded, no signup hassle |
| Bring your own key | OpenRouter, Anthropic, or OpenAI — any provider that speaks the OpenAI chat API |
export OPENROUTER_API_KEY="sk-or-v1-..."
python3 --version # 3.10+
uv --version
duckdb --version
opencode --version
Don't want to install Python/uv/opencode? Open the repo in a GitHub Codespace — everything is pre-installed, you just bring your key.
gh codespace create --repo motherduckdb/analytics-agent-duckdb-workshop
| Part | Topic | Time |
|---|---|---|
| 1 | DuckDB from Python | 10 min |
| 2 | Text-to-SQL & the agentic loop | 15 min |
| 3 | Context: skill file + COMMENT ON | 15 min |
| 4 | Deterministic tools with PydanticAI | 20 min |
| 5 | Distribute via FastMCP | 15 min |
Workshop materials for PyCon DE & PyData 2026.
Run Claude Code as an MCP server so any agent can delegate coding tasks to it
Browser automation using accessibility snapshots instead of screenshots
MCP server integration for DaVinci Resolve Studio
A Jetbrains IDE IntelliJ plugin aimed to provide coding agents the ability to leverage intelliJ's indexing of the codeba