Your data. Your AI. Your machine. Your control.
No subscriptions, no vendor lock-in, no cloud costs.
Want to skip the philosophy and just see what this does?
# 1. Install Nix (one-time setup)
curl --proto '=https' --tlsv1.2 -sSf -L https://install.determinate.systems/nix | sh -s -- install
# 2. Close and reopen your terminal, then:
curl -L https://pipulate.com/install.sh | bash
# 3. Launch it
cd ~/pipulate && nix developWhat you get: A local web app at http://localhost:5001 with step-by-step workflows, integrated AI chat, and a JupyterLab instance at http://localhost:8888. No cloud required.
Success looks like: Two browser tabs auto-open showing the Pipulate interface and JupyterLab.
Real examples of what people create with Pipulate:
- Keyword Research Pipeline: Input seed keywords → AI expansion → competition analysis → export spreadsheet
- Content Gap Analysis: Compare your site vs competitors → identify missing topics → prioritized content calendar
- Technical SEO Audits: Crawl site → check Core Web Vitals → generate action items → track fixes
- CSV Data Cleaning: Upload messy data → standardize formats → remove duplicates → validate results
- API Data Collection: Connect to APIs → fetch data in batches → transform to consistent format → store locally
- Report Generation: Combine multiple data sources → apply business rules → create branded reports
- Content Creation Pipeline: Research topics → generate outlines → write drafts → optimize for SEO
- Data Analysis Helper: Upload spreadsheet → AI suggests insights → create visualizations → export findings
Key advantage: Each workflow is a guided, step-by-step process that non-technical users can run repeatedly, while developers can customize the Python code behind each step.
╔═════════════════════════════════════════════════════════════════════════╗ Chip O'What?
║ 🎭 PIPULATE: LOCAL-FIRST AI SEO SOFTWARE & DIGITAL WORKSHOP ║ , O
║ ────────────────────────────────────────────────────────────────────── ║ \\ . O
║ ║ |\\/| o
║ 💬 Chip O'Theseus: "Welcome to your sovereign computing environment!" ║ / " '\
║ ║ . . .
║ 🌟 Where Python functions become HTML elements... ║ / ) |
║ 🌟 Where workflows preserve your creative process... ║ ' _.' |
║ 🌟 Where AI integrates locally and globally... ║ '-'/ \
╚═════════════════════════════════════════════════════════════════════════╝
The Challenge with Agentic AI: Powerful but unpredictable—you never know what you're gonna get.
The Pipulate Approach: Structured workflows that can leverage any AI—local, cloud, or hybrid—while maintaining complete visibility and control.
Think of it as putting guardrails on AI assistance. Instead of asking an AI to "figure it out," domain experts create step-by-step workflows that guide AI through proven processes. The AI gets structure, you get predictable results.
Pipulate: Your AI Swiss Army Knife: Whether you prefer local privacy, cloud power, or hybrid approaches, Pipulate provides the framework. Use local models for sensitive work, cloud APIs for heavy lifting, or both in the same workflow—your choice, your control.
🤖 AGENTIC MODE (Chaos) 🚂 AI ON RAILS (Pipulate)
═══════════════════════ ══════════════════════════
💥 GOES OFF 📊 LINEAR WORKFLOWS
HALF-COCKED! BY DOMAIN EXPERTS
│ │
▼ ▼
╔════════════════════╗ ┌─────────────────────┐
║ 🌪️ WILLY NILLY 🎲 ║ │ Step 1: Analyze▸ │
║ ║ VS │ Step 2: Process▸ │
║ Unpredictable ║ │ Step 3: Report▸ │
║ Results ║ │ Step 4: Export▸ │
╚════════════════════╝ └─────────────────────┘
│ │
▼ ▼
☁️ Trains Frontier Models 🏠 Keeps Domain Expertise Local
- 🖥️ Runs locally like a desktop app using modern web technologies
- 🐍 Simple linear workflow approach powered by HTMX for seamless interactivity
- 📓 Transforms Jupyter Notebooks into production-ready, step-by-step workflows
- 🤖 Integrated AI assistance using your own local models with complete privacy
- 🔧 Reproducible environments with Nix that work identically across all platforms
- 🎯 Perfect for SEO practitioners who want to turn technical expertise into guided, reusable workflows
Pipulate is a local-first, single-tenant desktop app framework featuring AI-assisted, step-by-step workflows. Designed to feel like an Electron app, it uniquely runs a full, reproducible Linux environment within a project folder using Nix, ensuring consistency across macOS, Linux, and Windows (via WSL).
🖥️ ELECTRON PATTERN 🌐 PIPULATE PATTERN
═══════════════════════ ═══════════════════════
┌─────────────────────────┐ ┌─────────────────────────┐
│ ELECTRON APP │ │ PIPULATE SETUP │
├─────────────────────────┤ ├─────────────────────────┤
│ ┌─────┐ ┌─────┐ ┌─────┐ │ │ ┌─────────────────────┐ │
│ │.exe │ │.dmg │ │.deb │ │ │ │ install.sh │ │
│ └─────┘ └─────┘ └─────┘ │ │ │ (Works on ALL OSes) │ │
│ Per-OS Installers │ │ └─────────────────────┘ │
└───────────┬─────────────┘ └───────────┬─────────────┘
│ │
▼ ▼
┌─────────────────────────┐ ┌─────────────────────────┐ This is new.
│ 📱 Native Window │ │ 🖥️ Terminal Console │ , O
│ ┌─────────────────┐ │ │ ┌─────────────────────┐ │ \\ . O
│ │ Web Browser │ │ │ │ nix develop │ │ |\\/| o
│ │ (Bundled) │ │ │ │ Starting servers... │ │ / " '\
│ │ ┌───────────┐ │ │ │ │ ✓ JupyterLab ready │ │ . . .
│ │ │ │ │ │ │ │ ✓ Pipulate ready │ │ / ) |
│ │ │ HTML │ │ │ │ └─────────────────────┘ │ ' _.' |
│ │ │ CSS │ │ │ + └─────────────────────────┘ '-'/ \
│ │ │ JS │ │ │ │
│ │ │ │ │ │ ▼
│ │ └───────────┘ │ │ ┌─────────────────────────┐
│ └─────────────────┘ │ │ 🌐 Regular Browser │
│ │ │ │ ┌─────────────────────┐ │
│ ▼ │ │ │ localhost:5001 │ │
│ ┌─────────────────┐ │ │ │ ┌─────────────────┐ │ │
│ │ Node.js │ │ │ │ │ Python/HTMX │ │ │
│ │ Runtime │ │ │ │ │ Workflows │ │ │
│ └─────────────────┘ │ │ │ │ Local AI │ │ │
└─────────────────────────┘ │ │ └─────────────────┘ │ │
│ └─────────────────────┘ │
✅ Feels like native app └─────────────────────────┘
❌ Multiple installers needed
❌ Platform-specific builds ✅ Universal installer
❌ Update distribution complexity ✅ Auto-updates via git
✅ Same experience all OSes
✅ Complete reproducibility
This isn't just another framework — it's a deliberate culmination of decades of tech evolution insights. Pipulate represents the "third act" approach to development (3rd time's the charm): choosing the most durable and lovable parts of the modern tech stack while rejecting the exhausting hamster wheel of framework churn.
If you are not an Empire builder and prefer craftsmanship over the rat race and want to build tools that last, then Pipulate may be for you. Pipulate embodies that philosophy — maximum creative freedom with minimum technical debt, recapturing that old Webmaster feeling.
🎡 THE FRAMEWORK CHURN CYCLE 🏰 COMPUTING SOVEREIGNTY
═══════════════════════════════ ═══════════════════════════
React → Vue → Angular → Svelte 🗽 Your Hardware
↑ ↓ 🗽 Your Data
Webpack ← Next.js ← Vite ← Remix 🗽 Your AI Choice
↑ ↓ 🗽 Your Code
Docker → K8s → Cloud → Serverless 🗽 Your Schedule
😵💫 Endless Learning 🗽 Your Hardware
💸 Migration Fatigue 🗽 Your Data
🔒 Platform Lock-in 🗽 Your AI Choice
📈 Growing Complexity 🗽 Your Code
🗽 Your Schedule
WITH
✨ Durable Tools:
🏃♂️ JUMP OFF THE WHEEL • Python (30+ years)
↓ • SQLite (built-in)
┌─────────────┐ • HTML/HTTP (timeless)
│ PIPULATE │ • Nix (reproducible)
│ Local-First │ • Cloud APIs (by choice)
│+ Any Cloud │
└─────────────┘ 🎯 Third Act Philosophy:
"Choose tools that will
outlast any framework"
-
Local-First Sovereignty: Your data, code, and AI run on your hardware by default—extending to cloud services when you choose. This guarantees privacy, eliminates surprise costs, and gives you complete control over when and how to scale.
-
WET Workflows, DRY Framework: Workflows are intentionally "WET" (explicit & step-by-step) for maximum clarity and customizability—perfectly mirroring Jupyter Notebooks. The underlying framework is "DRY" for efficiency.
-
The AI Advantage: AI makes WET practical. Tedious code maintenance and refactoring, once a weakness of WET, is now an area where AI excels, turning repetition into a strength for rapid, context-aware development. Our breakthrough Workflow Reconstruction System exemplifies this: intelligent AST-based transplantation of workflow components eliminates traditional OOP inheritance complexity while maintaining perfect code precision.
-
Radical Transparency ("Know EVERYTHING!"): We reject opaque enterprise patterns in favor of complete observability. State is managed in transparent SQLite tables and JSON blobs, making the entire system intuitive and debuggable. No black boxes, ever.
-
Reproducibility with Nix: Nix Flakes provide a perfect, reproducible Linux environment on macOS, Linux, and Windows (WSL), solving the "works on my machine" problem.
-
Future-Proof Stack: We rely on durable standards: Python, SQLite, HTML, and HTMX. This is a framework built to last.
- Empower End-Users (e.g., SEO Practitioners): Enable non-programmers to run powerful, AI-guided workflows (often ported from Jupyter Notebooks) without needing to interact with Python code directly.
- Serve Developers: Provide a simple, reproducible environment for building these workflows, leveraging integrated tooling like Jupyter, local LLMs, and a streamlined web framework.
Pipulate's WET philosophy extends to its technology choices, favoring simple, durable tools over complex abstractions:
The Cloud's popularity has been driven in part by developers not wanting to maintain multiple codebases or installers per OS. Thanks to Nix, that's all fixed.
- Nix Flakes: Manages dependencies and creates reproducible environments, ensuring consistency across developers and operating systems, with optional CUDA support. E.g. Is this a Linux-thing you're reading about here? A Windows thing? A Mac thing? The answer is: YES!!! All of the above — and if you've got cool acceleration hardware, it will even take advantage and utilize that too. Best of all worlds.
____ _ _ .--. ___________
| _ \ __ _ _ ____ _(_)_ __ (_)_ __ ,--./,-. |o_o | | | |
| | | |/ _` | '__\ \ /\ / / | '_ \| \ \/ / / # \ |:_/ | | | |
| |_| | (_| | | \ V V /| | | | | |> < | | // \ \ |_____|_____|
|____/ \__,_|_| \_/\_/ |_|_| |_|_/_/\_\ \ / (| | ) | | |
`._,._,' /'\_ _/`\ | | |
Solving the "Not on my machine" problem well. \___)=(___/ |_____|_____|
Nix serves as the "Noah's Ark" — preserving this perfect focus in a reproducible environment that works identically across all platforms. Once you've locked in the focus, it lasts for years or decades, all bottled up in infrastructure-as-code.
Pipulate integrates a carefully selected set of tools aligned with its philosophy:
- FastHTML: A Python web framework prioritizing simplicity. It generates HTML directly from Python objects (no template language like Jinja2) and minimizes JavaScript by design, working closely with HTMX. It's distinct from API-focused frameworks like FastAPI. The Python function-naming is the HTML-template language.
🏛️ ORIGINAL LAMP STACK (2000s) 🚀 NEW LAMP STACK (2025)
═══════════════════════════════ ═══════════════════════════
┌─────────────────────────────┐ ┌─────────────────────────────┐
│ 🐧 L: Linux │ │ 🐧 L: Linux + Nix │
│ Single OS, manual setup │ │ Reproducible everywhere │
├─────────────────────────────┤ ├─────────────────────────────┤
│ 🌐 A: Apache │ │ ⚡ A: ASGI │
│ Static config, restarts │ │ Async, hot reload │
├─────────────────────────────┤ ├─────────────────────────────┤
│ 🗄️ M: MySQL │ │ 📊 M: MiniDataAPI │
│ Complex queries, joins │ │ Python-native simplicity│
├─────────────────────────────┤ ├─────────────────────────────┤
│ 🔧 P: PHP │ │ 🐍 P: Python + FastHTML │
│ Mix of HTML/logic │ │ + HTMX │
└─────────────────────────────┘ └─────────────────────────────┘
│ │
▼ ▼
┌─────────────────────┐ ┌─────────────────────────┐
│ 🏢 Enterprise │ │ 🏠 Local-First │
│ Complexity │ │ Sovereignty │
│ │ │ │
│ • Multi-server │ │ • Single machine │
│ • Load balancers │ │ • Integrated AI │
│ • Database clusters │ VS │ • SQLite simplicity │
│ • DevOps overhead │ │ • Nix reproducibility │
│ • Cloud lock-in │ │ • Flexible deployment │
└─────────────────────┘ └─────────────────────────┘
🎯 One person understands 🎯 One person controls
part of the system the entire system
The original LAMP stack was beautiful in its simplicity — one person could understand and manage the whole stack. But it got bloated with enterprise patterns, microservices, and distributed complexity.
Pipulate brings back that "one person, full stack" philosophy with modern tools:
- Linux + Nix: Reproducible environments across all platforms
- ASGI: Modern async server interface, future-proofed for performance
- MiniDataAPI: Universal SQL simplifier close to Python's core data structures
- Python + FastHTML + HTMX: The new web development paradigm
This isn't just simpler — it's more powerful, giving you complete environment reproducibility, local AI integration, server-side state management, and future-proofed skills.
Pipulate's technology choices form aligned lenses that focus ideas from abstraction to actualization. Each lens must be ground and polished without misaligning the focus:
Universal Translator of Abstractions clarify into implementations
Spoken Language to Code by each lens being simple and transparent.
Idea --> Lens 1 --> Lens 2 --> Lens 3 -> Lens 4 -> Lens 5 -> Lens 6
-----> ,--.
---> ,' `.---------> ,--.
--> / \------> ,' `.-------> ,--. ,-.
o -> / Linux \----> / http \----> ,'_hx `.--->,' `. ,-.
/|\ ( HARDWARE )--> ( PROTOCOL )--> ( LINGUA )->( UI/UX )->(APP)->(git)
/ \ -> \ Nix /----> \ html /----> `..py ,'--->`. ,' `-'
--> \ /------> `. ,'-------> `--' `-' And so on
---> `. ,'---------> `--' AI Help
-----> `--' AI Help
AI Help
We keep lenses minimal, their material either thoroughly pre-trained into the model (Python 3.x, HTMX, etc.) or able to be included in the prompt and easily held in the context window. We've trimmed the cruft — the lens flashes and burrs, and all unnecessary extra lenses (Angular, React, Vue, etc.)
HARDWARE:
install.sh: Published on Pipulate.com to initiate magic cookie install
flake.nix: Nix IaC creating a normalized Linux subsystem on any host OS
PROTOCOL:
http: Uvicorn fast Asynchronous Server Gateway Interface (ASGI) web server
html: Uvicorn talks to Python Starlette using anyio & httpx libraries
LINGUA:
htmx: /assets/htmx.js JavaScript library to eliminate most need for JavaScript
Python: .venv/bin/python3.12 latest version AIs are well trained on
UI/UX:
browser: Obviously, but I guess it needs to be said. Like a looser Electron.
fasthtml: /assets/fasthtml.js for FT Components, Python functions as templating
APP:
app: Flask-style Uvicorn factory instance instantiated by FastHTML fast_app
db: Dict-like DB providing transparent server-side state (server cookies)
pipulate: Pipeline state management, like db but with JSON blob for workflowsIn lens manufacturing, "flashes" are excess material that squeeze out of molds — unwanted projections that must be ground off. Steve Jobs famously did this twice: adopting Gorilla Glass (grinding off plastic flashes) and rejecting Flash Player (grinding off software bloat).
Pipulate continues this tradition:
- FastHTML: Grinds off Jinja2 template complexity
- HTMX: Grinds off virtual DOM overhead
- Local AI: Enables privacy by default, cloud power when desired
- SQLite: Grinds off enterprise database complexity
The result: clean, focused tools that do their job without unnecessary cruft.
This is not your father's Python web framework. HTMX changes everything — a marriage made in heaven between Python and the Web, finally turning Python into a first-class citizen for web development. In many use cases such as this one, Python is even preferable to JavaScript in the way it blends Python's formidable ecosystem of packages with workflows.
The revolution isn't just another framework — it's eliminating the template layer entirely:
🍶 FLASK ERA 🚀 FASTAPI ERA 🌐 FASTHTML ERA
═══════════════ ═══════════════ ══════════════════
┌─────────────┐ ┌─────────────┐ ┌─────────────┐
│ Python │ │ Python │ │ Python │
│ Functions │ │ Functions │ │ Functions │
└──────┬──────┘ └──────┬──────┘ └──────┬──────┘
│ │ │
▼ ▼ ▼
┌─────────────┐ ┌─────────────┐ ┌─────────────┐
│ Jinja2 │ │ Pydantic │ │ HTMX │◄─ Over-the-wire
│ Templates │ │ Models │ │ Fragments │ HTML targeting
└──────┬──────┘ └──────┬──────┘ └──────┬──────┘ DOM elements
│ │ │
▼ ▼ ▼
┌─────────────┐ ┌─────────────┐ ┌─────────────┐
│ HTML │ │ JSON │ │ HTML │
│ Response │ │ Response │ │ Elements │
└─────────────┘ └─────────────┘ └─────────────┘
│ │ │
▼ ▼ ▼
🌐 Full Page Reload 📱 Frontend Framework 🎯 DOM Element Updates
(React/Vue/Angular) def Div() = <div>
def Button() = <button>
Template files needed JSON ↔ HTML conversion Python functions ARE
Separate languages Client-side complexity the template language!
The FastHTML Breakthrough: Python function names directly become HTML elements, eliminating templates and making the server the single source of truth for UI state.
-
HTMX: Enables dynamic, interactive UIs directly in HTML via attributes, minimizing the need for custom JavaScript. Pipulate uses it for server-rendered HTML updates — over the wire HTML-fragments targeting elements of the DOM directly instead of fragile, performance-reducing, framework-dependent JSON. THIS is where you jump off the tech-churn hamsterwheel and future-proof yourself.
-
MiniDataAPI: A lightweight layer for interacting with SQLite and other databases. Uses Python dictionaries for schema definition, promoting type safety without the complexity of traditional ORMs — effectively future-proofing your SQL. You lose fancy join capabilities but in exchange get the Python dict interface as your main persistent database API forever-forward, enabiling instant swapability between SQLite and PostgreSQL (for example).
-
Ollama: Facilitates running LLMs locally, enabling in-app chat, workflow guidance, and future automation capabilities while ensuring privacy and avoiding API costs. Your local AI (Chip O'Theseus) learns & grows with you, hopping from hardware to hardware as you upgrade — like a genie in a hermitcrab shell. And if that weren't kooky enough — it knows how to make MCP-calls!!! That's right, your friendly localhost AI Chip O'Theseus is also an MCP client! Your linear workflows ain't so linear anymore when a single-step can be: "Go out and do whatever."
Pipulate isn't anti-cloud—it's pro-choice. Each workflow step can choose the best tool for the job:
- Step 1: Use local AI for sensitive data analysis (privacy-first)
- Step 2: Call OpenAI's API for advanced reasoning (cloud power)
- Step 3: Process results locally and save to SQLite (data sovereignty)
- Step 4: Use Anthropic's API for final review (frontier capabilities)
This is the Swiss Army knife approach: Local by default, cloud by choice, with complete visibility into what's happening at each step. Whether you're processing confidential client data (local) or need cutting-edge AI capabilities (cloud), Pipulate gives you the framework to do both seamlessly.
- SQLite & Jupyter Notebooks: Foundational tools for data persistence and the workflow development process (porting from notebooks to Pipulate workflows). SQLite is built into Python and really all things — the get-out-of-tech-liability free card you didn't know you had. And a full JupyterLab instance is installed side-by-side with Pipulate sharing the same Python
.venvvirtual environment, which is also shared with your preferred AI code editor (Cursor, Windsurf, VSCode, Zed) so... well... uhm, there are no words for when 3 different portals-to-Python share the same environment. You can do such stupid AI-tricks as letting your local LLM and a frontier cloud model inhabit the same body (Pipulate) — controlling web browsers together and stuff.
We offer two installation paths that lead to the exact same robust, Nix-managed environment. Choose the path that best fits your experience level and preferences.
┌────────────────────────────┐
│ New User on macOS │
└─────────────┬──────────────┘
│
┌───────────────────────┴───────────────────────┐
│ │
▼ ▼
┌──────────────────────────────────┐ ┌───────────────────────────────────────────┐
│ PATH 1: Recommended for Everyone │ │ PATH 2: Alternative for Python Developers │
└──────────────────────────────────┘ └───────────────────────────────────────────┘
│ │
"I want the simplest, most "I prefer managing my command-line
direct way to get this running." tools with standard Python utilities."
│ │
▼ ▼
1. `curl ... [nix]` 1. `brew install pipx` (If needed)
2. `curl ... [pipulate]` 2. `pipx install pipulate`
3. `pipulate install`
│ │
└───────────────┐ ┌───────────────┘
│ │
▼ ▼
┌────────────────────────────┐
│ Nix-Managed Pipulate │
│ Environment │
└────────────────────────────┘
||
(Identical
Result)
This is the fastest and most universal way to install Pipulate. It has the fewest dependencies and works on any modern Mac, Linux system, or Windows with WSL.
📦 Your Machine 🔧 Add Foundation 🚀 Complete Environment
Today with Nix Ready to Go!
┌─────────────┐ ┌─────────────┐ ┌─────────────┐
│ Sad Computer│ Step 1 │ 🏗️ Nix │ Step 2 │ 🎯 Pipulate │
│ Without │ ──────────► │ Foundation │ ──────────► │ + AI + │
│ Nix😢 │ │ Installed │ │ Jupyter │
└─────────────┘ └─────────────┘ └─────────────┘
│
Step 3 │
▼
┌─────────────┐
│ 🌐 Browser │
│ Opens │
│Automatically│
└─────────────┘
Simple as 1-2-3! No Docker, no build steps, works with or without cloud services.
Everything runs locally with complete flexibility and control.
Step 1: Install Nix (One-Time Setup)
If you don't have it already, install the Nix package manager. It's the system that makes Pipulate's reproducible environment possible.
curl --proto '=https' --tlsv1.2 -sSf -L https://install.determinate.systems/nix | sh -s -- installImportant: After the Nix installation finishes, you must close and reopen your Terminal window.
Step 2: Run the Pipulate Installer
Now, run the universal install script. You can give your project a custom name, too.
# To install with a custom name like "Botifython"
curl -L https://pipulate.com/install.sh | bash -s Botifython
# Or, to install with the default name "pipulate"
curl -L https://pipulate.com/install.sh | bashStep 3: Launch Pipulate
Navigate into your new project directory and launch the environment with nix develop.
# cd into the directory you just created
cd ~/Botifython
# Launch Pipulate
nix developThat's it! The server and JupyterLab will start, and the application will open in your browser.
Running It Again:
- You can just forcibly exit out of that Terminal it's running from.
- Open a new Terminal, and once again:
cd ~/Botifython
nix developThe Big Reset (If Necessary):
Things sometimes go wrong. This is how you do a full Pipulate reset. This will also delete anything you downloaded with Pipulate. Adjust custom install name to what you used.
rm -rf ~/Botifython
curl -L https://pipulate.com/install.sh | bash -s Botifython
cd ~/Botifython
nix developWait for BOTH TABS to auto-open in your browser.
Common Issues & Solutions:
| Problem | Solution |
|---|---|
nix: command not found |
You didn't restart your terminal after Nix installation |
| Browser doesn't open automatically | Manually visit http://localhost:5001 and http://localhost:8888 |
Permission denied errors |
Make sure you can write to ~/pipulate directory |
| Port conflicts | Kill processes on ports 5001/8888: lsof -ti:5001 | xargs kill -9 |
| Nix build fails | Clear Nix cache: nix-collect-garbage then retry |
System Requirements:
- macOS: 10.15+ (Intel/Apple Silicon)
- Linux: Any modern distribution with curl
- Windows: WSL2 with Ubuntu 20.04+
- RAM: 4GB minimum, 8GB recommended
- Disk: 2GB for installation + data storage
- Network: Internet connection for initial setup only
If you are a developer comfortable with tools like Homebrew and pipx, you can use our PyPI package as a gateway to the same robust installation.
Step 1: Install pipx
pipx is a tool for safely installing Python command-line applications. If you don't have it, you can install it with Homebrew.
brew install pipxStep 2: Install the Pipulate CLI
Use pipx to install the pipulate command-line tool. This will not cause conflicts with your system Python.
pipx install pipulateStep 3: Run the Installer
Use the command you just installed to set up the full Pipulate application.
pipulate installThis will trigger the same universal installation process, resulting in the exact same robust, Nix-managed environment. To run it in the future, just type pipulate run.
These few commands:
- ✅ Updates to the latest version automatically
- ✅ Starts JupyterLab and the Pipulate server
- ✅ Opens web interfaces in your browser
- ✅ Provides a complete, reproducible development environment
That's it! You now have a local-first development environment with AI integration, installed via your preferred Python toolchain.
Here's what happens behind the scenes during the "magic cookie" installation:
User runs install.sh (via curl) Nix Flake Activation & Transformation
┌──────────────────────────────┐ ┌────────────────────────────────────────────┐
│ 1. Download install.sh │ │ 5. User runs 'nix develop' │
│ 2. Download ZIP from GitHub │ │ 6. Flake detects non-git directory │
│ 3. Extract ZIP to ~/AppName │ │ 7. Flake clones repo to temp dir │
│ 4. Download ROT13 SSH key │ │ 8. Preserves app_name.txt, .ssh, .venv │
│ to .ssh/rot │ │ 9. Moves git repo into place │
└─────────────┬────────────────┘ │10. Sets up SSH key for git │
│ │11. Transforms into git repo │
▼ │12. Enables auto-update via git pull │
┌─────────────────────────────────────────────────────────────────────────────┐
│ Result: Fully functional, auto-updating, git-based Pipulate installation │
└─────────────────────────────────────────────────────────────────────────────┘
Are you a Developer or an End User? Chef or Customer? Understanding your audience is crucial for effective development. Pipulate serves two distinct but complementary audiences, much like a restaurant serves both chefs and customers
┌──────────────────────────────────────────────────────────┐
│ The Restaurant │
│ ┌──────────────────┐ ┌──────────────────┐ │
│ │ Kitchen (Dev) │ │ Dining Room │ │
│ │ │ │ (End Users) │ │
│ │ │ │ │ │
│ │ 👨🍳 Sous Chef │───recipes───►│ 🍽️ Customers │ │
│ │ 👩🍳 Head Chef │ │ 🏢 Restaurateur │ │
│ │ │ │ │ │
│ │ "How do we make │ │ "I want the best │ │
│ │ pasta you've │ │ pasta I've ever │ │
│ │ never had?" │ │ had in my life" │ │
│ └──────────────────┘ └──────────────────┘ │
└──────────────────────────────────────────────────────────┘
- 🔧 Workflow Creators: Build and customize AI-assisted workflows
- 📓 Jupyter Porters: Convert notebook experiments into guided applications
- 🔍 Technical SEOs: Create sophisticated, reusable SEO processes
- ⚙️ System Administrators: Deploy consistent environments across teams
What Chefs Get:
- 🎛️ Complete control over the "recipe" (workflow logic)
- 🔄 Reproducible development environment via Nix
- 🏗️ Simple architecture that's easy to understand and modify
- 🧰 Integrated tooling (Jupyter, local LLM, SQLite)
- 📈 SEO Practitioners: Run powerful workflows without coding
- ✍️ Content Creators: Use AI-assisted processes for optimization
- 📊 Marketing Teams: Execute consistent SEO strategies
- 🏢 Business Owners: Access enterprise-level SEO capabilities
What Customers Get:
- 🚶♂️ Guided, step-by-step workflow experiences
- 🤖 AI assistance at every step
- 🙈 No need to see or understand the underlying code
- 🎯 Consistent, repeatable results
Just as a chef talks about knife techniques while a diner just wants amazing pasta, Pipulate separates the complexity of creation from the simplicity of consumption. Developers craft the workflows, end-users enjoy the results.
After installation succeeds, here's what to expect:
-
Two browser tabs open automatically:
localhost:5001- Pipulate web interface with navigation menulocalhost:8888- JupyterLab for development/experimentation
-
In the Pipulate interface:
- Left sidebar with workflow plugins (Introduction, Profiles, etc.)
- Main area showing step-by-step workflow interface
- Right panel with integrated AI chat (Chip O'Theseus)
-
Terminal shows:
🚀 Starting Pipulate servers... ✓ FastHTML server ready at http://localhost:5001 ✓ JupyterLab ready at http://localhost:8888 ✓ Local AI ready for chat assistance
🔍 If you're an SEO practitioner:
- Click "Introduction" in the left menu for a guided tour
- Try the built-in workflows to see the step-by-step pattern
- Use the AI chat to ask "How do I create a keyword research workflow?"
👨💻 If you're a developer:
- Open JupyterLab tab and run the introduction notebook
- Check out
plugins/010_introduction.pyto see workflow code structure - Try creating a simple workflow:
python helpers/workflow/create_workflow.py
🤖 If you're an AI assistant:
- Focus on the Quick Reference Card above
- Study the Critical Implementation Patterns section
- Review
mcp_tools.pyfor MCP protocol capabilities
🆕 If you're just exploring:
- Click through the left menu items to see different workflow types
- Ask the AI chat: "What can I build with Pipulate?"
- Try the Introduction workflow to see the step-by-step experience
Pipulate is built on a radical philosophy that challenges programming orthodoxy: WET (Write Everything Twice) is better than DRY (Don't Repeat Yourself) when you have AI to help manage it.
At every scale of reality, we see the same pattern: "lumps of stuff" with APIs that enable interaction. Quarks combine into atoms, atoms into molecules, cells into organisms, individuals into societies. Each level requires the right granularity of interface — not so abstract that you lose control, not so granular that you drown in complexity.
This is the 80/20 rule of existence: Handle 80% of interactions gracefully with 20% of the API surface, then handle edge cases as needed. Pipulate applies this principle to code architecture.
The tech industry suffers from "hamster wheel syndrome" — constantly breaking APIs that force migration cycles. React (20+ versions), Node (frequent breaking changes), Angular (complete rewrites). This isn't progress; it's planned obsolescence.
Pipulate chooses durable foundations:
- Linux Kernel: Version 6 in 30 years
- Python: Version 3 in 30 years
- HTML: Version 5 and stable
- HTTP: Version 3 and backward compatible
These are the "laws of physics" for software — stable APIs that enable compound growth rather than constant rebuilding.
Traditional development follows DRY principles, creating abstract, complex systems that are hard to understand and modify. But the world has changed:
- 🔬 Jupyter Notebooks promote explicit, literate programming
- 🤖 AI assistants excel at managing repetitive code
- 🏠 Local-first architectures prioritize clarity over enterprise complexity
- 🎯 Right Granularity: WET provides the perfect abstraction level for human AND AI comprehension
________________________________
- Like Notebooks / \
- Linear Workflows | It runs proprietary private AI |
- Local & Cloud-free | Workflows from your Local PC?! |
- Chip O'Theseus included \________________________________/
()
HARDWARE PLATFORM LOCAL BROWSER O , Chip O'Theseus
_______________________ __________ _______ o \\ .
| | / Pipulate \Jupyter\__ |\\/|
| Windows, Mac or Linux | | __________________ | See! / " '\ - Radical transparency
| _____ ___ | | | App Name Menu⚙️| |<- - - - -. . . - MCP tool-call control
| _/ Nix \____\_____ | | |------------------| | / ) | - Browser as bot's body
| | | | | | Workflow | Local | | ' _.' |
| | Pipulate <---------> -Step #1 | AI🤖 | | '-'/ \
|__| localhost:5001 |_| | | -Step #2 | Chat | | What, no Docker?
| (AI on Rails🚂) | | | -Step #3 | Help▸ | | What, no React?
|__________________| | |__________|_______| | What, no Cloud?
|______________________|
WET workflows are:
- 🔍 Observable: See exactly what's happening at every step
- 🔧 Customizable: Modify workflows without breaking abstractions
- 🤖 AI-Friendly: Clear code that AI assistants can easily understand and maintain
- 🚀 Future-Proof: Built on durable web standards that won't become obsolete
Nix Environment Activation: Always run nix develop from the ~/pipulate directory before running any project commands (python server.py, pip install, etc.) in a new terminal. This ensures you are using the correct dependencies defined in flake.nix.
Interactive vs. Quiet Shell:
Standard Shell: nix develop (or nix develop .#default) runs the startup script (run-script defined in flake.nix) with welcome messages and service startup. Ideal for general use.
Quiet Shell: nix develop .#quiet activates the Nix environment without running the full startup script or launching services automatically. It only sets up paths and installs pip requirements. Use this for:
- Running specific commands without starting the servers (e.g.,
nix develop .#quiet --command python -c "import pandas"). - Debugging or interacting with AI assistants where verbose startup output is undesirable.
- Manually running
run-serverorrun-jupyter(scripts placed in.venv/binby theshellHook).
Dependencies: System-level dependencies (Python version, libraries like gcc, zlib) are managed by flake.nix. Python package dependencies are managed by pip using requirements.txt within the Nix-provided environment.
Source of Truth: The flake.nix file is the definitive source for the development environment setup.
Pipulate features a distinct architecture designed for its local-first, simple, and observable nature.
This diagram illustrates the high-level components and their interactions:
┌─────────────┐ Like Electron, but full Linux subsystem
│ Browser │ in a folder for macOS and Windows (WSL)
└─────┬───────┘
│ HTTP/WS
▼
┌───────────────────────────────────────┐
│ Nix Flake Shell │ - In-app LLM (where it belongs)
│ ┌───────────────┐ ┌──────────────┐ │ - 100% reproducible
│ │ FastHTML │ │ Ollama │ │ - 100% local
│ │ HTMX App │ │ Local LLM │ │ - 100% multi-OS
│ └───────┬───────┘ └──────────────┘ │
│ │ │
│ ┌─────▼─────┐ ┌────────────┐ │
│ │MiniDataAPI│◄───►│ SQLite DB │ │
│ └───────────┘ └────────────┘ │
└───────────────────────────────────────┘
This complete, self-contained environment runs identically on any operating system, providing the foundation for all Pipulate workflows and AI interactions.
Jupyter Notebooks run alongside the FastHTML server, allowing developers to prototype workflows in a familiar environment before porting them to Pipulate's step-based interface for end-users. The same Python virtual environment (.venv) is shared, and ad-hoc package installation is supported. If you're using Cursor, VSCode or Windsurf, set your Ctrl+Shift+P "Python: Set Interpreter" to "Enter Interpreter Path" ./pipulate/.venv/bin/python. You might have to adjust based on the folder you use as your workspace. But then you'll have a Python environment unified between Cursor, JupyterLab and Pipulate.
┌──────────────────┐ ┌──────────────────┐
│ Jupyter Lab │ │ FastHTML │
│ Notebooks │ │ Server │
│ ┌──────────┐ │ │ ┌──────────┐ │
│ │ Cell 1 │ │ │ │ Step 1 │ │
│ │ │ │--->│ │ │ │
│ └──────────┘ │ │ └──────────┘ │
│ ┌──────────┐ │ │ ┌──────────┐ │
│ │ Cell 2 │ │ │ │ Step 2 │ │
│ │ │ │--->│ │ │ │
│ └──────────┘ │ │ └──────────┘ │
│ localhost:8888 │ │ localhost:5001 │
└──────────────────┘ └──────────────────┘
Pipulate manages all state server-side within the local environment (think local-server cookies), with optional cloud integration as needed. This approach offers:
- Privacy & Control: Data stays local by default, cloud integration when you choose.
- Full Resource Access: Utilize local CPU/GPU freely for intensive tasks, plus cloud APIs for heavy lifting.
- Simplicity: Eliminates complexities of multi-tenancy while supporting both local and cloud workflows.
- Observability: State changes (via DictLikeDB/JSON) are transparent and easily logged (AI greps it there).
This detailed view shows how Pipulate's local-first architecture eliminates common web development complexities:
┌───────────────────────────────┐ # Benefits of Local-First Simplicity
│ Web Browser │
│ │ - No mysterious client-side state
│ ┌────────────────────┐ │ - No full-stack framework churn
│ │ Server Console │ │ - No complex ORM or SQL layers
│ │ & Web Logs │ │ - No external message queues
│ └─────────┬──────────┘ │ - No build step required
│ ▼ │ - Direct, observable state changes
│ ┌─────────────────────┐ │
│ │ Server-Side State │ │ - Conceptually like local-server-side cookies
│ │ DictLikeDB + JSON ◄──────── Enables the "Know EVERYTHING!" philosophy
│ └─────────────────────┘ │ - AI greps logs/server.log to see app state!
└───────────────────────────────┘
The UI is constructed primarily with server-rendered HTML fragments delivered via HTMX. This minimizes client-side JavaScript complexity.
- FastHTML generates HTML components directly from Python.
- HTMX handles partial page updates based on user interactions, requesting new HTML snippets from the server.
- WebSockets and Server-Sent Events (SSE) provide real-time updates (e.g., for chat, live development reloading).
HTMX+Python enables a world-class
Python front-end Web Development environment.
┌─────────────────────┐
│ Navigation Bar │ - No template language (like Jinja2)
├─────────┬───────────┤ - HTML elements are Python functions
Simple Python back-end │ Main │ Chat │ - Minimal custom JavaScript / CSS
HTMX "paints" HTML into │ Area │ Interface │ - No React/Vue/Angular overhead
the DOM on demand ───────► │ │ │ - No "build" process like Svelte
└─────────┴───────────┘ - No virtual DOM, JSX, Redux, etc.
With such minimal surface area the AI code assistant knows everything. LLMs are either pre-trained on the stable, infrequently revved libraries used (Python 3.12, HTMX, or it's all small enough to fit in a 1-shot prompt — yes, the whole core code-base fits in one Gemini Web UI form submit.
Designed for porting notebook-style processes, workflows are sequences of steps where the state is managed explicitly at each stage and stored persistently (typically as a JSON blob in the pipeline table).
- Resumable & Interrupt-Safe: Because each step's completion is recorded, workflows can be stopped and resumed.
- Explicit State Flow: Data typically passes from one step's output (
donefield) to the next via thetransformfunction, simplifying debugging. Patterned on Unix pipes. - Good Training Data: The structured input/output of each step creates valuable data for potentially fine-tuning models.
- Proprietary Friendly: Excellent for proprietary domain-experts and fields (competing academic, finances) who resist letting their data flow onto the Web for general AI training.
┌─────────┐ ┌─────────┐ ┌─────────┐ - Fully customizable steps
│ Step 01 │─piped─►│ Step 02 │─piped─►│ Step 03 │ - Interruption-safe & resumable
└─────────┘ └─────────┘ └─────────┘ - Easily ported from Notebooks
│ │ │ - One DB record per workflow run
▼ ▼ ▼ - Everything stays on your machine
State Saved State Saved Finalized? - Magnitudes simpler than celery
The key insight: Pipulate workflows use a run_all_cells() pattern that directly mirrors Jupyter's "Run All Cells" command. This creates an immediate mental model — each workflow step is like a notebook cell, and the system automatically progresses through them top-to-bottom, just like running all cells in a notebook.
📓 JUPYTER NOTEBOOK 🌐 PIPULATE WORKFLOW
═══════════════════ ══════════════════════
[ ] Cell 1: Import data ┌─────────────────────┐
│ │ Step 1: Data Input │
▼ └──────────┬──────────┘
[▶] Cell 2: Process data │ hx_trigger="load"
│ ▼
▼ ┌─────────────────────┐
[ ] Cell 3: Generate report │ Step 2: Processing │
│ └──────────┬──────────┘
▼ │ hx_trigger="load"
[ ] Cell 4: Export results ▼
┌─────────────────────┐
🎯 "Run All Cells" Button ═══► │ Step 3: Export │
Executes top-to-bottom └─────────────────────┘
Same mental model, same execution flow!
But with persistent state, a web UI and
not having to look at the Python code 🚫🐍.
Integration with a local Ollama instance provides AI capabilities without external API calls:
- Privacy: Prompts and responses stay local.
- Cost-Effective: No per-token charges; run continuously using local resources.
- Streaming Support: Real-time interaction via WebSockets.
- Bounded Context: Manages conversation history effectively.
- App State Awareness: Grepping your server log reveals full application state.
- Tool Calling: Local LLM is an MCP client with a growing list of abilities
- Workflow assistance
- Browser automation
- Debugging
┌──────────────────┐
│ Local Ollama │ - No API keys needed
│ Server │ - Completely private processing
└────────┬─────────┘
│
│ Streaming via WebSocket
▼
┌──────────────────┐
│ Pipulate App │ - Monitors WS for MCP tool-call commands
│(WebSocket Client)│ - Parses responses in real-time
└────────┬─────────┘
│
│ In-memory or DB backed
▼
┌──────────────────┐
│ Bounded │ - Manages context window (~128k)
│ Chat History │ - Enables RAG / tool integration
└──────────────────┘
Nix Flakes ensure a consistent environment across Linux, macOS, and Windows (via WSL), optionally leveraging CUDA GPUs if detected.
┌──────────────────┐
│ Linux / macOS │ - Write code once, run anywhere
│ Windows (WSL) │ - Consistent dev environment via Nix
└────────┬─────────┘ - As if Homebrew but across all OSes
│
│ Nix manages dependencies
▼
┌──────────────────┐
│ CUDA Support │ - Auto-detects NVIDIA GPU w/ CUDA
│ (if present) │ - Uses GPU for LLM acceleration
└──────────────────┘ - Falls back to CPU if no CUDA
The application interface is organized into distinct areas:
┌─────────────────────────────┐
│ Navigation ◄── Search, Profiles,
├───────────────┬─────────────┤ Apps, Settings
│ │ │
Workflow, ──► Main Area │ Chat │
App UI │ (Pipeline) │ Interface ◄── LLM Interaction
│ │ │
└─────────────────────────────┘
Critical for AI assistants: All UI components use semantic IDs and comprehensive ARIA labeling for accessibility and automation.
🏠 home (Root Component)
├── 📦 create_outer_container()
│ ├── 🧭 create_nav_group() [id='nav-group', role='navigation', aria-label='Main navigation']
│ │ ├── 🔍 nav_search_container [role='search', aria-label='Plugin search']
│ │ │ ├── Input [id='nav-plugin-search', role='searchbox', aria-label='Search plugins']
│ │ │ └── Div [id='search-results-dropdown', role='listbox', aria-label='Search results']
│ │ ├── 👤 create_profile_menu() [id='profile-dropdown-menu', aria-label='Profile management']
│ │ │ ├── Summary [id='profile-id', aria-label='Profile selection menu']
│ │ │ └── Ul [role='menu', aria-label='Profile options', aria-labelledby='profile-id']
│ │ ├── ⚡ create_app_menu() [id='app-dropdown-menu', aria-label='Application selection']
│ │ │ ├── Summary [id='app-id', aria-label='Application menu']
│ │ │ └── Ul [role='menu', aria-label='Application options', aria-labelledby='app-id']
│ │ ├── 🌍 create_env_menu() [id='env-dropdown-menu', data-testid='environment-dropdown-menu']
│ │ │ ├── Summary [id='env-id', aria-label='Environment selection menu']
│ │ │ └── Ul [role='menu', aria-label='Environment options', aria-labelledby='env-id']
│ │ └── ⚙️ poke_section [id='poke-dropdown-menu']
│ │ ├── Summary [id='poke-summary']
│ │ └── Div [id='nav-flyout-panel']
│ ├── 📱 main-grid
│ │ ├── 📋 create_grid_left() [id='grid-left-content'] → Workflow Steps/Cells Display
│ │ │ ├── content_to_render (Dynamic workflow content)
│ │ │ └── scroll_to_top [id='scroll-to-top-link']
│ │ └── 🤖 create_chat_interface() [id='chat-interface', role='complementary', aria-label='AI Assistant Chat']
│ │ ├── H2 [APP_NAME + ' Chatbot']
│ │ ├── Div [id='msg-list', role='log', aria-label='Chat conversation', aria-live='polite']
│ │ └── Form [role='form', aria-label='Chat input form']
│ │ ├── Textarea [id='msg', role='textbox', aria-label='Chat message input', aria-multiline='true']
│ │ ├── Button [id='send-btn', aria-label='Send message to AI assistant']
│ │ └── Button [id='stop-btn', aria-label='Stop AI response streaming']
│ └── 🔧 HTMX Refresh Listeners
│ ├── Div [id='profile-menu-refresh-listener', hx_target='#profile-dropdown-menu']
│ └── Div [id='app-menu-refresh-listener', hx_target='#app-dropdown-menu']
Navigation Updates:
#profile-dropdown-menu- Profile menu refresh target#app-dropdown-menu- App menu refresh target#search-results-dropdown- Live search results#nav-flyout-panel- Settings flyout panel
Content Areas:
#grid-left-content- Main workflow display area#msg-list- Chat conversation historybody- Full page navigation refreshes
Interactive Elements:
#nav-plugin-search- Real-time plugin search (300ms delay)#send-btn/#stop-btn- Chat control buttons#scroll-to-top-link- Scroll navigation aid
This structure enables AI assistants to programmatically interact with all UI components using semantic selectors and ARIA landmarks.
.
├── .cursor/ # Bootstraps Radical Transparency (teaches AI to fish)
│ └── rules/ # Framework rules (01_CRITICAL_PATTERNS.mdc, etc.)
├── .venv/ # Common Python environment for FastHTML, Jupyter & Cursor
├── browser_automation/ # Selenium browser control & DOM capture
│ ├── looking_at/ # Current browser DOM state for AI visibility
│ └── *.py # Google search automation examples
├── cli.py # Command line interface for Pipulate operations
├── common.py # Base Class for DRY CRUD plugin app inheritance (todo)
├── data/
│ └── data.db # AI-accessible SQLite for application state (server cookies)
├── downloads/ # Default location for workflow outputs (e.g., CSVs)
├── helpers/
│ ├── botify
│ │ └── botify_api.ipynb # Git managed massive example notebook, produces docs
│ ├── workflow # Workflow workshop, lots of tools that make WET DRY
│ │ └── create_workflow.py # Example of what might be found there
│ └── prompt_foo.py # Bundles XML code payloads for massive 1-shot AI prompts
├── logs/
│ ├── server-1.log # N-rotations of server log per run per config
│ └── server.log # The server log of most recent run, contains app state
├── /assets/ # JS, CSS, images, icons
├── plugins/ # Workflow plugins (010_introduction.py, 400_trifecta.py, etc.)
├── pyproject.toml # Python packaging configuration and metadata
├── training/ # Markdown files for AI context/prompts
├── vulture_whitelist.py # Code analysis whitelist for unused code detection
├── flake.nix # Infrastructure as Code & all system-versions for AI
├── LICENSE # It's MIT
├── install.sh # "Magic cookie" installation script (curl | sh)
├── mcp_tools.py # MCP protocol tools - the AI assistant interface
├── notebook_introduction_local.ipynb # Editable (non-auto-updating) copy of hello.ipynb
├── README.md # This file
├── requirements.txt # Python dependencies (managed by Nix)
└── server.py # Main application entry point
These patterns are essential for LLMs working with Pipulate and are frequently missed:
📝 AUTO-KEY GENERATION FLOW
┌─────────────┐ POST ┌─────────────┐ HX-Refresh ┌─────────────┐
│ Empty Form │ ──────────► │ Server │ ──────────────► │ Page Reload │
│ Submit ⏎ │ /init │ Response │ Header │ Trigger │
└─────────────┘ └─────────────┘ └─────────────┘
▲ │
│ ▼
┌─────────────┐ ┌─────────────┐ ┌─────────────┐
│ User Hits │ ◄─────────── │ Auto-Key │ ◄───────────── │ landing() │
│ Enter Again │ Ready! │ Populated │ Generates │ Method │
└─────────────┘ └─────────────┘ └─────────────┘
When a user hits Enter on an empty key field, this specific sequence occurs:
- Form Submission: POSTs to
/{APP_NAME}/initwith emptypipeline_id - Server Response: The
initmethod MUST return anHX-Refreshresponse:if not user_input: from starlette.responses import Response response = Response('') response.headers['HX-Refresh'] = 'true' return response
- Page Reload: HTMX triggers a full page reload
- Auto-Key Population: The
landing()method callspip.generate_pipeline_key(self)to populate the input field - User Interaction: User hits Enter again to start the workflow
Pipulate uses HTMX-driven step progression powered by the brilliantly named run_all_cells() method:
- Initial Trigger: After
init, therun_all_cells()method initializes the workflow just like Jupyter's "Run All Cells" - Perfect Mental Model: The method name creates immediate understanding — workflows execute top-to-bottom like notebook cells
- Step Handlers: Each step has GET (display) and POST (submit) handlers
- Automatic Progression: Completed steps trigger next step with
hx_trigger="load" - State Persistence: Each step stores data in pipeline state
- Pedagogical Brilliance: The naming makes the system instantly intuitive for developers and AI assistants
Example: The run_all_cells() Pattern in Action
# ✅ CORRECT: Use the run_all_cells() method for workflow initialization
async def init(self, request):
"""Initialize workflow using the run_all_cells pattern"""
return pip.run_all_cells(app_name, steps)
# ❌ ANTI-PATTERN: Manual placeholder creation
async def init(self, request):
"""Manual approach — harder to understand and maintain"""
first_step_id = steps[0].id
return Div(
Div(id=first_step_id, hx_get=f'/{app_name}/{first_step_id}', hx_trigger='load'),
id=f"{app_name}-container"
)The run_all_cells() method encapsulates the workflow initialization pattern and creates an immediate mental connection to Jupyter notebooks.
📂 FILENAME vs APP_NAME DISTINCTION
┌─────────────────────────────────────────────────────────────┐
│ CRITICAL SEPARATION │
├─────────────────────────────────────────────────────────────┤
│ │
│ 📁 FILENAME: 200_workflow_genesis.py │
│ ├── 🌐 Determines public URL: /workflow_genesis │
│ └── 📊 Controls menu order: 200 │
│ │
│ 🏷️ APP_NAME: "workflow_genesis_internal" │
│ ├── 💾 Database table identifier │
│ ├── 🔒 MUST REMAIN STABLE (data integrity) │
│ └── 🚫 NEVER change after deployment │
│ │
│ ⚠️ DANGER: Changing APP_NAME = Orphaned Data │
└─────────────────────────────────────────────────────────────┘
Critical for data integrity:
- Filename (e.g.,
200_workflow_genesis.py): Determines public URL endpoint and menu ordering - APP_NAME Constant (e.g.,
APP_NAME = "workflow_genesis_internal"): Internal identifier that MUST REMAIN STABLE
- State stored as JSON blobs in pipeline table
- Accessed via
pip.get_step_data()andpip.set_step_data() - All state changes are transparent and observable
📁 PLUGIN DISCOVERY SYSTEM
plugins/
├── 010_introduction.py ✅ Registered as "introduction" (menu order: 1)
├── 020_profiles.py ✅ Registered as "profiles" (menu order: 2)
├── hello_flow (Copy).py ❌ SKIPPED - Contains "()"
├── xx_experimental.py ❌ SKIPPED - "xx_" prefix
└── 200_workflow_genesis.py ✅ Registered as "workflow_genesis" (menu order: 20)
📊 AUTO-REGISTRATION RULES:
✅ Numeric prefix → Menu ordering + stripped for internal name
❌ Parentheses "()" → Development copies, skipped
❌ "xx_" prefix → Work-in-progress, skipped
🔧 Must have: landing() method + name attributes
💉 Auto dependency injection via __init__ signature
- Files in
plugins/directory are auto-discovered - Numeric prefixes control menu ordering
- Classes must have
landingmethod and name attributes - Automatic dependency injection based on
__init__signature
Pipulate includes sophisticated helper scripts for workflow development:
Creates new workflows from templates:
python create_workflow.py workflow.py MyWorkflow my_workflow \
"My Workflow" "Welcome message" "Training prompt" \
--template trifecta --forceAdds steps to existing workflows:
python splice_workflow_step.py workflow.py --position top
python splice_workflow_step.py workflow.py --position bottom🏗️ WORKFLOW TEMPLATE SYSTEM
┌─────────────────┐ ┌─────────────────┐
│ BLANK TEMPLATE │ │TRIFECTA TEMPLATE│
├─────────────────┤ ├─────────────────┤
│ ┌─────────────┐ │ │ ┌─────────────┐ │
│ │ Step 1 │ │ │ │ Step 1 │ │
│ │ (Minimal) │ │ │ │ (Input) │ │
│ └─────────────┘ │ VS │ └──────┬──────┘ │
│ │ │ │ │
│ Quick Start │ │ ▼ │
│ Single Purpose │ │ ┌─────────────┐ │
└─────────────────┘ │ │ Step 2 │ │
│ │ (Process) │ │
create_workflow.py │ └──────┬──────┘ │
--template blank │ │ │
│ ▼ │
│ ┌─────────────┐ │
│ │ Step 3 │ │
│ │ (Output) │ │
│ └─────────────┘ │
│ │
│ Full Pattern │
│ Complete Flow │
└─────────────────┘
create_workflow.py
--template trifecta
blank: Minimal workflow with one steptrifecta: Three-step workflow pattern- Automatic method generation and insertion
The Revolutionary Alternative to OOP Inheritance: Atomic transplantation of workflow components using intelligent pattern matching and AST precision.
🧬 WORKFLOW RECONSTRUCTION: ATOMIC TRANSPLANTATION
═════════════════════════════════════════════════════════════════════════
OLD WORKFLOW WORKFLOW UPDATED WORKFLOW
(Atomic Source) RECONSTRUCTOR (Incremental Gen)
┌─────────────────┐ ┌─────────────────┐ ┌─────────────────┐
│ 🧬 Components: │ AST │ 🎯 Pattern │ AST │ ✨ Generated: │
│ │ ───► │ Matching │ ───► │ │
│ ┌─────────────┐ │ │ │ │ ┌─────────────┐ │
│ │step_params* │ │ │ Bundle Type 1: │ │ │step_params* │ │ ✅
│ │step_optim* │ │ │ Auto-Registered │ │ │step_optim* │ │ ✅
│ │parameter* │ │ │ Methods │ │ │parameter* │ │ ✅
│ └─────────────┘ │ │ │ │ └─────────────┘ │
│ │ │ Bundle Type 2: │ │ │
│ ┌─────────────┐ │ │ Custom Routes │ │ ┌─────────────┐ │
│ │_process │ │ │ (_process, │ │ │_process │ │ ✅
│ │preview │ │ │ preview) │ │ │preview │ │ ✅
│ └─────────────┘ │ │ │ │ └─────────────┘ │
└─────────────────┘ └─────────────────┘ └─────────────────┘
🔄 COMPLETE LIFECYCLE: Test → Validate → Production → Cleanup
--suffix 5 --target new_name --target same_name git status
────────── ──────────────── ────────────────── ──────────
param_buster5 advanced_params param_buster (in-place) (shows cruft)
(safe testing) (new workflow) (git history preserved) (clean up!)
🎯 WHY IT WORKS: Lightning in a Bottle
┌─────────────────────────────────────────────────────────────────────────┐
│ ✨ Pattern Matching: No manual markers needed │
│ 🔧 AST Precision: Syntactically perfect code generation │
│ 🎭 Inheritance Alternative: Compose without complex super() chains │
│ 🧪 Safe Testing: Incremental validation without production risk │
│ 📚 Git Continuity: In-place updates preserve development history │
│ 🧹 Systematic Cleanup: Prevents file cruft accumulation │
└─────────────────────────────────────────────────────────────────────────┘
workflow_reconstructor.py --template botify_trifecta
--source parameter_buster
--suffix 5
The System That Eliminates Bootstrap Paradox:
- Atomic Sources: Battle-tested workflows become component libraries
- Pattern Matching: Intelligent detection via
_process,previewpatterns - AST Transplantation: Surgical precision without syntax errors
- Complete Lifecycle: Development → Testing → Production → Cleanup
# Development workflow
cd ~/pipulate && nix develop # Start Pipulate
nix develop .#quiet # Start without auto-services
python server.py # Manual server start
git pull && nix develop # Update to latest
# Create new workflows
python helpers/workflow/create_workflow.py my_workflow.py MyClass my_internal_name
python helpers/workflow/splice_workflow_step.py my_workflow.py --position top
# Plugin naming conventions
010_my_plugin.py # Active plugin (menu order 1)
xx_my_plugin.py # Disabled during development
my_plugin (Copy).py # Ignored development copy- Pipulate App:
http://localhost:5001 - JupyterLab:
http://localhost:8888 - Local AI Chat: Built into the Pipulate interface
- Logs:
tail -f logs/server.logfor debugging
# Auto-key generation flow
if not user_input:
response = Response('')
response.headers['HX-Refresh'] = 'true'
return response
# Workflow initialization
return pip.run_all_cells(app_name, steps)
# State management
data = pip.get_step_data(step_id)
pip.set_step_data(step_id, updated_data)plugins/ # Your workflows (auto-discovered)
├── 010_introduction.py # Menu order 1
├── xx_draft.py # Disabled (xx_ prefix)
└── draft (Copy).py # Ignored (parentheses)
mcp_tools.py # AI assistant interface
common.py # Base classes for workflows
browser_automation/ # Selenium automation tools
logs/server.log # Debug everything here
data/data.db # SQLite application state
🚨 LLM IMPLEMENTATION MISTAKE PREVENTION
┌────────────────────────────────────────────────────────────┐
│ COMMON PITFALLS │
├────────────────────────────────────────────────────────────┤
│ ❌ Missing HX-Refresh │ ✅ if not user_input: │
│ Response │ response.headers['HX- │
│ │ Refresh'] = 'true' │
├────────────────────────────┼───────────────────────────────┤
│ ❌ Wrong Key Generation │ ✅ pip.generate_pipeline_ │
│ Method │ key(self) │
├────────────────────────────┼───────────────────────────────┤
│ ❌ Broken Chain Reaction │ ✅ hx_trigger="load" → │
│ Pattern │ Automatic progression │
├────────────────────────────┼───────────────────────────────┤
│ ❌ APP_NAME Changes │ ✅ NEVER modify after │
│ (Data Orphaning) │ deployment │
└────────────────────────────────────────────────────────────┘
LLMs frequently make these errors:
- Missing HX-Refresh Response: Forgetting to return the refresh response for empty keys
- Incorrect Key Generation: Not using
pip.generate_pipeline_key(self)properly - Missing Cursor Positioning: Forgetting the
_onfocusattribute for user experience - Wrong Route Handling: Not understanding the difference between landing page and init routes
- State Inconsistency: Not properly handling the key generation and storage flow
- APP_NAME Changes: Modifying APP_NAME after deployment, orphaning existing data
- Chain Reaction Breaks: Not properly implementing the HTMX step progression pattern
These "speedbumps" reinforce Pipulate's core philosophy:
- Local vs. Enterprise Mindset: Embrace local-first simplicity. Avoid patterns designed for distributed, multi-tenant systems.
- JSON State Management (Workflows): Keep workflow state in self-contained steps within a single JSON blob per run. Avoid complex state machines or external step tracking.
- Database (MiniDataAPI): Use the simple schema definition and access patterns provided. Avoid heavy ORMs.
- Workflow Pattern: Ensure workflows are linear and state is explicitly passed or saved at each step. Avoid complex async task chaining that obscures state.
- UI Rendering Pattern: Generate HTML directly from Python components via FastHTML. Avoid template engines.
- WebSocket Pattern: Use the dedicated
Chatclass for managing LLM interactions. Avoid raw WebSocket handling elsewhere. - Workflow Progression Pattern: Workflows use an explicit chain reaction pattern with
hx_trigger="load"to manage step progression. This pattern must be preserved exactly as implemented. See the workflow documentation for details.
- Monitoring: A file system watchdog monitors code changes. Valid changes trigger an automatic, monitored server restart via Uvicorn, facilitating live development.
┌─────────────┐ ┌──────────────┐
│ File System │ Changes │ AST Syntax │ Checks Code
│ Watchdog │ Detects │ Checker │ Validity
└──────┬──────┘ └───────┬──────┘
│ Valid Change │
▼ ▼
┌───────────────────────────┐ ┌──────────┐
│ Uvicorn Server │◄─── │ Reload │ Triggers Restart
│ (Handles HTTP, WS, SSE) │ │ Process │
└───────────────────────────┘ └──────────┘
The system provides comprehensive debugging and state inspection capabilities through MCP tools and real-time monitoring:
📊 PIPELINE STATE INSPECTOR
├─── 🔍 Discovering active workflows...
├─── ⚡ Found 3 running processes
├─── 🎯 Step 2/5: Processing data transformations
└─── ✨ Ready for next interaction!
🤖 MCP TOOL CALLS
├─── 📡 Connecting to Botify API...
├─── 🔐 Authentication successful
├─── 📊 Fetching schema (4,449 fields discovered!)
└─── 💾 Caching results for lightning-fast access
Pipulate represents more than technology choices — it's a statement about computing sovereignty and the future of human-AI collaboration.
How do you "lift yourself up by your bootstraps"? The answer: you're not a closed system. Humans interact with reality through interfaces, just like software components. We are "human apps with reality APIs."
This pattern repeats at every scale — from quantum fields to societies. The art lies in choosing the right granularity for these interfaces. Too abstract, and you lose control. Too granular, and you drown in complexity.
The most successful APIs are durable — they become the "laws of physics" for their domain. In reality: the laws of physics, chemistry, biology. In tech: logic gates, operating systems, protocols like HTTP and HTML.
Pipulate chooses these durable APIs deliberately, creating a future-proof foundation for AI-assisted development that will outlast the current framework churn.
This documentation serves multiple audiences: humans seeking immediate productivity, and AI systems learning to collaborate with human creativity. The explicit, observable nature of Pipulate workflows creates perfect training data for the next generation of AI assistants.
Pipulate offers a unique approach to building local, AI-assisted applications, prioritizing simplicity, user control, and reproducibility over conventional scaling patterns. By understanding its core concepts and embracing its philosophy, developers can create powerful workflows, and end-users can benefit from AI guidance with complete flexibility in deployment options.
Remember the guiding principles:
- Keep it simple.
- Support local-first with cloud integration options.
- Embrace the unconventional while remaining practical.
- Choose durable foundations that work with any approach.
- Build for both human creativity and AI collaboration—local or cloud.
The Bottom Line: Pipulate doesn't reject the modern AI ecosystem—it provides a structured foundation that works with any AI service. Whether you're using Claude via API, ChatGPT for reasoning, or local models for privacy, Pipulate gives you the workflow framework to orchestrate them all effectively. It's not about choosing sides in the AI wars—it's about having the right tool for any job.
The repository includes not only polished plugins but also experimental scripts and notebooks under development (e.g., in the root directory or marked with xx_ prefix in plugin directories). These represent ongoing work and exploration.
- Numeric Prefixes: Files like
workflows/10_hello_flow.pyare registered ashello_flow(number stripped for internal name, used for menu order). - Parentheses Skip: Files with
()in the name (e.g.,hello_flow (Copy).py) are skipped – useful for temporary copies during development. xx_Prefix Skip: Files prefixed withxx_(e.g.,xx_experimental_flow.py) are skipped – useful for keeping unfinished work in the plugin directories without activating it.
- Copy: Copy a template to
my_flow (Copy).py. - Modify: Develop your workflow. It won't auto-register yet.
- Test: Rename to
xx_my_flow.py. The server should auto-reload. Test thoroughly. - Deploy: Rename to
##_my_flow.pyto assign menu order and activate.
Use git mv for simple renames (like xx_ to numbered prefix) to preserve history. Document more complex renames in commit messages.
git mv workflows/xx_my_flow.py workflows/##_my_flow.py
git commit -m "Feat: Promote workflow xx_my_flow.py to ##_my_flow.py"This README serves as the upstream source of truth for all Pipulate documentation across GitHub, Pipulate.com, and the built-in app documentation. Changes made here automatically cascade to all other documentation surfaces.
🌊 THE UPSTREAM TRUTH CASCADE
═══════════════════════════════════════════════════════════════
📜 Source Code Reality (The Ultimate Truth)
│
▼
📄 README.md (Single Source of Truth)
├─ ASCII Art Blocks (Visual Truth)
├─ HTML Comment Keys (Metadata)
└─ 80-Hyphen Pagination (Structure)
│
┌──────────────┼──────────────────┐
▼ ▼ ▼
🌐 GitHub Page 📚 Pipulate.com 🔧 Built-in Docs
(Auto-display) (Jekyll Build) (Live Integration)
│ │ │
▼ ▼ ▼
📊 Screenshots 🎬 Demos 🧪 Tests
(Future) (Future) (Future)
How it works:
- ASCII Art Blocks: Visual diagrams are automatically extracted and distributed to other documentation files
- HTML Comment Keys: Headlines marked with
<!-- key: identifier -->serve as reference anchors - 80-Hyphen Pagination: Section dividers enable automatic document structuring
- Automatic Synchronization: Running
python helpers/docs_sync/sync_ascii_art.pyupdates all documentation
This creates "ASCII art peer pressure" — when visual diagrams change, they compel surrounding text to be updated for consistency, ensuring documentation accuracy across the entire ecosystem.
Core & Workflow Enhancements:
- Dev, Test, and Prod database switching
- Saving source HTML and rendered DOM of any URL
- Botify data export CSV save (incorporating robust polling)
- Full web form field support (textarea, dropdown, checkboxes, radio buttons)
- Generic support for Anywidgets
- Utility for deleting garbage tables from plugin experimentation
AI / LLM Integration:
- LLM inspection of any local data object (RAG-style functionality)
- Various memory types for LLM context (vector embedding, graph, key/val-store)
- Enabling the local LLM to be an MCP Client
Automation & External Interaction:
- MCP Server for automated web browsing and similar tasks
THEMES
- Okaidia ocodia 1.77KB
LANGUAGES
- CSS1.71KB
- Markup + HTML + XML + SVG + MathML + SSML + Atom + RSS4.64KB
- C-like0.83KB
- JavaScript6.18KB
- Bash + Shell + Shell zeitgeist87 8.96KB
- Diff uranusjr 1.33KB
- JSON + Web App Manifest CupOfTea696 0.58KB
- JSON5 RunDevelopment 0.52KB
- JSONP RunDevelopment 0.23KB
- Liquid cinhtau 2.56KB
- Lua Golmote 0.74KB
- Markdown Golmote 10.43KB
- Markup templating
- Mermaid RunDevelopment 3.03KB
- Nix Golmote 1.47KB
- Python multipetros 2.45KB
- Regex RunDevelopment 2.33KB
- YAML hason 3.11KB
PLUGINS
- Line Highlight11.66KB
- Line Numbers kuba-kubula 7.23KB
- Toolbar mAAdhaTTah 5.63KB
Contributions are welcome! Please adhere to the project's core philosophy:
- Maintain Local-First Simplicity (No multi-tenant patterns, complex ORMs, heavy client-side state).
- Respect Server-Side State (Use DictLikeDB/JSON for workflows, MiniDataAPI for CRUD).
- Preserve the Workflow Pipeline Pattern (Keep steps linear, state explicit).
- Honor Integrated Features (Don't disrupt core LLM/Jupyter integration unless enhancing local goals).
This project is licensed under the MIT License. See the LICENSE file for details.
Background Articles: Mike Levin, AI SEO in NYC
Enhanced Documentation: Pipulate AI SEO Software
On GitHub: Pipulate on GitHub