An open API service indexing awesome lists of open source software.

https://github.com/raphaelbleier/dataiku-ai-context

Dataiku DSS documentation bundles + Claude Code subagents for every major AI coding tool
https://github.com/raphaelbleier/dataiku-ai-context

Last synced: 2 months ago
JSON representation

Dataiku DSS documentation bundles + Claude Code subagents for every major AI coding tool

Awesome Lists containing this project

README

          

# Dataiku AI Context

> Full Dataiku DSS documentation + Claude Code subagents — installable into every major AI coding tool in one command.

[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE)
[![Agents](https://img.shields.io/badge/agents-6-blue)](#claude-code-agents)
[![Doc bundles](https://img.shields.io/badge/doc%20bundles-87-green)](#documentation-bundles)
[![Tools supported](https://img.shields.io/badge/tools-7-purple)](#supported-ai-tools)

## What's included

| | |
|---|---|
| **6 Claude Code agents** | Domain-specialized subagents for Dataiku development |
| **87 doc bundles** | Full DSS + Developer docs as markdown, topic-organized |
| **7 tool installers** | Claude Code, Cursor, Copilot, Windsurf, Aider, Continue, Cline |

Docs scraped from [doc.dataiku.com](https://doc.dataiku.com/dss/latest/) and [developer.dataiku.com](https://developer.dataiku.com/latest/) — 1755 pages bundled into ~400 KB topic files.

---

## Quick install

```bash
git clone https://github.com/raphaelbleier/dataiku-ai-context.git
cd dataiku-ai-context
./install.sh
```

The interactive installer lets you choose:
- **Claude Code agents** (global or project-local)
- **Doc bundles** for your AI coding tool
- **Both**

---

## Claude Code agents

Six specialized subagents, auto-loaded by Claude Code from `~/.claude/agents/`:

| Agent | When to use |
|-------|------------|
| `dataiku-developer` | General DSS tasks: Flow, datasets, recipes, project structure |
| `dataiku-data-engineer` | Pipelines, Spark/SQL, scenarios, automation, metrics |
| `dataiku-ml-engineer` | Visual ML, AutoML, MLflow, model deployment, monitoring |
| `dataiku-genai-engineer` | LLM Mesh, AI agents, RAG, Knowledge Banks, GenAI recipes |
| `dataiku-api-developer` | Python API, REST API, API Node, Project Deployer |
| `dataiku-admin` | Installation, users, connections, code envs, security, ops |

### Install agents only

```bash
# Global (all projects)
./install.sh
# → choose "Claude Code agents" → "Global"

# Or copy manually
cp agents/*.md ~/.claude/agents/

# Or project-local
mkdir -p .claude/agents
cp agents/*.md .claude/agents/
```

---

## Documentation bundles

87 markdown files covering every DSS topic, ready to load as context in any AI tool.

```
docs/
├── INDEX.md ← start here
├── concepts.md
├── python-api.md
├── dev-concepts-examples.md ← Python API cookbook
├── machine-learning.md
├── generative-ai-llm.md
├── ai-agents.md
├── code-recipes.md
├── flow.md
├── data-preparation.md
├── automation-scenarios.md
├── mlops.md
├── deployment.md
└── ... (87 total)
```

---

## Supported AI tools

| Tool | What gets created |
|------|------------------|
| **Claude Code** | `.claude/dataiku/` + `CLAUDE.md` snippet |
| **Cursor** | `.cursor/dataiku/` + `.cursor/rules/dataiku.mdc` |
| **GitHub Copilot** | `.github/dataiku/` + `.github/copilot-instructions.md` |
| **Windsurf** | `.windsurf/dataiku/` + `.windsurfrules` |
| **Aider** | `dataiku-docs/` + `DATAIKU_DOCS.md` |
| **Continue.dev** | `.continue/dataiku/` + `.continue/config.json` |
| **Cline** | `.cline/dataiku/` + `.clinerules` |

### Install docs for a specific tool

```bash
# Interactive
./install.sh
# → choose "AI tool doc bundles" → pick your tool

# Python CLI (more control)
python tools/dataiku-docs.py install cursor --target /path/to/your/project
python tools/dataiku-docs.py install all --target /path/to/your/project
python tools/dataiku-docs.py list
```

---

## Usage in Claude Code

Once agents are installed, Claude Code automatically delegates Dataiku tasks to the right agent:

```
# These automatically invoke the matching subagent:
"Help me write a Python recipe to join two datasets" → dataiku-developer
"Set up MLflow tracking for my model" → dataiku-ml-engineer
"Build a RAG pipeline with Dataiku Knowledge Banks" → dataiku-genai-engineer
"Automate this pipeline with a scenario" → dataiku-data-engineer
"Create an API endpoint for my model" → dataiku-api-developer
"Add a new user and configure LDAP" → dataiku-admin
```

Reference doc bundles directly:

```
@.claude/dataiku/python-api.md
@.claude/dataiku/generative-ai-llm.md
@.claude/dataiku/INDEX.md
```

---

## Keeping docs up to date

Re-scrape and re-bundle when Dataiku releases a new DSS version:

```bash
python tools/scrape_docs.py # re-scrapes ~1755 pages
python tools/dataiku-docs.py bundle # rebuilds bundles
```

---

## Repository structure

```
dataiku-ai-context/
├── agents/ # Claude Code subagent definitions
│ ├── dataiku-developer.md
│ ├── dataiku-ml-engineer.md
│ ├── dataiku-data-engineer.md
│ ├── dataiku-genai-engineer.md
│ ├── dataiku-api-developer.md
│ └── dataiku-admin.md
├── docs/ # 87 bundled documentation files
│ ├── INDEX.md
│ └── *.md
├── tools/ # Maintainer scripts
│ ├── scrape_docs.py # Re-scrapes Dataiku docs
│ └── dataiku-docs.py # Bundle + install CLI
├── install.sh # Interactive installer
├── LICENSE
└── README.md
```

---

## Contributing

See [CONTRIBUTING.md](CONTRIBUTING.md).

PRs welcome for:
- New/improved agent prompts
- Bug fixes in the installer
- Support for additional AI tools

---

## License

MIT — see [LICENSE](LICENSE).

---

*Docs sourced from [doc.dataiku.com](https://doc.dataiku.com/dss/latest/) and [developer.dataiku.com](https://developer.dataiku.com/latest/) under Dataiku's documentation terms.*