📖 Live Docs · ⚡ CLI Repository · 🐛 Report Issue · 💬 Discussions
One contract. Every cloud. Zero lock-in.
Define your data product in a single YAML file — schema, quality rules, lineage, governance, SLAs — then deploy it to GCP BigQuery, AWS Athena, Snowflake, or DuckDB with one command. No rewrites. No vendor lock-in. Ever.
Data teams are drowning in glue code. Every cloud has its own APIs, its own deploy patterns, its own way of doing quality checks. You end up maintaining N copies of the same logic across N platforms — and it still breaks on Friday at 5pm.
Fluid Forge kills that complexity with 44 purpose-built CLI commands and a declarative contract system.
| Problem | Fluid Forge Solution |
|---|---|
| Cloud-specific deploy scripts everywhere | One declarative YAML contract that runs on any provider |
| Schema drift breaks pipelines silently | Built-in validate + verify catches drift before and after deploy |
| No visibility into data quality | test with quality gates, SLA checks & anomaly detection — in the contract |
| Orchestration is hand-wired spaghetti | export generates Airflow, Dagster, Prefect & Step Functions from your contract |
| Governance is an afterthought | policy-check → policy-compile → policy-apply — governance-as-code, compiled to native IAM |
| Switching clouds means rewriting everything | Swap one line (provider: gcp → provider: snowflake) and redeploy |
| AI/LLM access to data is ungoverned | agentPolicy — declarative boundaries on which models can consume your data |
pip install fluid-forge
fluid init my-project --quickstart
fluid apply contract.fluid.yaml --yesThat's it. Your data product is live — running on DuckDB locally, no cloud account needed. When you're ready for production, change one line and push. Same contract, same guarantees, infinite scale.
# contract.fluid.yaml — your entire data product in one file
fluidVersion: "0.7.1"
kind: DataProduct
id: "gold.crypto.bitcoin_tracker_v1"
name: "Bitcoin Price Tracker"
metadata:
owner:
team: data-engineering
email: data-team@company.com
domain: crypto
tags: [real-time, pricing]
exposes:
- exposeId: bitcoin_prices
title: "Bitcoin Hourly Prices"
kind: table
contract:
schema:
- name: timestamp
type: timestamp
required: true
- name: price_usd
type: float64
description: "Bitcoin price in USD"
binding:
platform: local # swap to gcp, aws, or snowflake anytime
format: duckdb_table
location:
database: crypto_data
table: bitcoin_prices
dataQuality:
rules:
- type: not_null
field: price_usd
- type: freshness
max_age: "1h"
governance:
accessPolicy:
- role: analyst
access: read
- role: data-engineer
access: admin| 🏠 Local (DuckDB) Develop & test instantly No cloud account needed |
☁️ GCP (BigQuery + GCS) Production analytics Cloud Composer, Pub/Sub, IAM |
🔶 AWS (S3 + Athena + Glue) Serverless queries on your data lake |
❄️ Snowflake Enterprise warehouse Snowpark & dbt integration |
Plus: Build your own providers with the Provider SDK — Databricks, Azure, Postgres, anything.
Fluid Forge isn't just apply. It's a complete data product lifecycle toolkit:
| Category | Commands | What It Does |
|---|---|---|
| Declare & Deploy | init · validate · plan · apply · execute |
Build, validate, and deploy data products from YAML contracts |
| Test & Verify | test · verify · contract-tests · diff |
Live resource validation, schema compatibility, drift detection |
| Orchestration | export · generate-airflow · scaffold-composer · scaffold-ci |
Auto-generate Airflow DAGs, Dagster graphs, Prefect flows, CI/CD pipelines |
| Governance | policy-check · policy-compile · policy-apply |
Validate policies, compile to native IAM, and enforce — all from the contract |
| Visualization | viz-graph · viz-plan · preview |
Lineage diagrams (SVG/PNG/HTML), execution plan visualization |
| Publishing | publish · export-opds · odcs · datamesh-manager |
Register in catalogs, export to OPDS/ODCS, push to Data Mesh Manager |
| AI & Blueprints | forge --mode copilot · forge --mode agent · blueprint · marketplace |
Adaptive AI-assisted creation, spec-backed domain agents, blueprint templates, marketplace discovery |
| Config & Admin | context · providers · doctor · auth · wizard |
Provider management, diagnostics, interactive onboarding |
Run
fluid doctorto verify your setup, orfluid wizardfor interactive onboarding.
Governance isn't a plugin — it's a first-class citizen in every contract:
- Column-level sensitivity — Tag PII, classify data at the field level
- Access policies — RBAC rules that compile to native cloud IAM (BigQuery, Snowflake, AWS)
- Data sovereignty — Jurisdiction and residency enforcement baked into the contract
- Agentic governance — Control which AI/LLM models can access your data and for what purposes
- Quality gates — Anomaly detection, SLA thresholds, and freshness checks that block bad deploys
fluid policy-check contract.fluid.yaml # Validate policies
fluid policy-compile contract.fluid.yaml # Compile to native IAM JSON
fluid policy-apply contract.fluid.yaml # Enforce on infrastructureEverything you need to go from first install to production-grade data products:
| Section | Description |
|---|---|
| Getting Started | Install & build your first data product in under 2 minutes |
| Walkthroughs | Step-by-step guides — Local, GCP, Airflow, Jenkins CI/CD |
| CLI Reference | Full command reference for all 45 top-level CLI commands |
| Providers | Deep dives into GCP, AWS, Snowflake, Local & Custom providers |
| Advanced | Blueprints, governance, Airflow integration, AI-powered agents |
| Vision & Roadmap | Where we're headed and how to shape the future |
git clone https://github.com/Agentics-Rising/forge_docs.git
cd forge_docs
npm install
npm run docs:dev # Dev server at http://localhost:8080Build for production:
npm run docs:build # Output → docs/.vuepress/dist/
npm run docs:preview # Preview the production builddocs/
├── README.md # Home page (live site landing)
├── vision.md # Philosophy & roadmap
├── contributing.md # How to contribute
├── RELEASE_NOTES_0.7.1.md # Latest release notes
├── getting-started/ # Installation & first steps
├── walkthrough/ # Step-by-step guides (Local, GCP, Airflow, Jenkins)
├── cli/ # CLI command reference
├── providers/ # Provider docs (GCP, AWS, Snowflake, Local, Custom)
├── advanced/ # Blueprints, governance, Airflow integration, AI agents
└── .vuepress/
├── config.ts # VuePress configuration & navigation
└── public/ # Static assets (logo, favicon)
Pushes to main automatically build and deploy to GitHub Pages via the workflow in .github/workflows/deploy-docs.yml.
Live docs: https://agentics-rising.github.io/forge_docs/
Note: To enable GitHub Pages, go to your repository Settings → Pages and set the source to GitHub Actions.
We welcome contributions! Whether it's fixing a typo, improving an explanation, or adding a new guide:
- Fork this repository
- Create a branch (
git checkout -b docs/my-improvement) - Install dependencies with
npm ci - Make your changes — the dev server hot-reloads on save
- Run
npm run docs:build - Open a Pull Request
If the docs change accompanies a CLI change, include the related forge-cli PR in the docs PR description. Docs-only PRs are welcome too.
For detailed guidelines, see CONTRIBUTING.md and our Code of Conduct.
| Repository | Description |
|---|---|
forge-cli |
The Fluid Forge CLI — the core engine |
forge-cli-sdk |
SDK for building custom providers |
Copyright 2025-2026 Agentics Transformation Pty Ltd.
Licensed under the Apache License, Version 2.0. See LICENSE for the full license text and NOTICE for attribution details.
Built with VuePress 2 · Powered by Fluid Forge
Declarative Data Products for Modern Data Teams
Proudly developed by dustlabs.co.za
