dataiku-ai-context

Dataiku AI Context

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

What's included

Docs scraped from doc.dataiku.com and developer.dataiku.com — 1755 pages bundled into ~400 KB topic files.

Quick install

Linux / macOS

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

Windows (PowerShell)

git clone https://github.com/raphaelbleier/dataiku-ai-context.git
cd dataiku-ai-context
.\install.ps1

One-liners (no clone required)

# Linux / macOS
curl -sL https://raw.githubusercontent.com/raphaelbleier/dataiku-ai-context/main/install.sh | bash

# Windows
iex (irm https://raw.githubusercontent.com/raphaelbleier/dataiku-ai-context/main/install.ps1)

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/:

Install agents only

# Linux / macOS — interactive
./install.sh
# → choose "Claude Code agents" → "Global"

# Linux / macOS — manual
cp agents/*.md ~/.claude/agents/

# Linux / macOS — project-local
mkdir -p .claude/agents && cp agents/*.md .claude/agents/

# Windows — interactive
.\install.ps1

# Windows — manual (global)
Copy-Item agents\*.md "$HOME\.claude\agents\"

# Windows — project-local
New-Item -ItemType Directory -Force .claude\agents | Out-Null
Copy-Item 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

Each tool gets 6 domain-specific rule files matching the Claude Code agents:

The installer copies both doc bundles and rule files in one step.

Install docs for a specific tool

# 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:

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 (6 files)
│   └── dataiku-*.md
├── rules/                  # Agent rules for other AI coding tools
│   ├── cursor/             # .mdc files for Cursor
│   ├── copilot/            # .instructions.md for GitHub Copilot
│   ├── windsurf/           # .md files for Windsurf
│   ├── aider/              # .md files for Aider
│   ├── continue/           # .md files for Continue.dev
│   └── cline/              # .md files for Cline
├── 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 (Linux/macOS)
├── install.ps1             # Interactive installer (Windows)
├── LICENSE
└── README.md

Contributing

See CONTRIBUTING.md.

PRs welcome for:

• New/improved agent prompts
• Bug fixes in the installer
• Support for additional AI tools

License

MIT — see LICENSE.

Docs sourced from doc.dataiku.com and developer.dataiku.com under Dataiku's documentation terms.