Pipelex is an agent-first tool for building repeatable AI workflows. Using Pipelex, agents can transform natural language requirements into production-ready AI workflows in minutes. The Pipelex language (PLX) is an open standard.

Quick links: Website • Docs • Cookbook (examples) • Starter template • VS Code extension: vscode-pipelex

💡 Recommended editor: Install vscode-pipelex to edit .plx pipeline files. It works with VS Code, Cursor, Windsurf, BlackboxAI, and most VS Code forks.

• pipelex — Core library & PLX language Define, run, and compose pipelines (sequences/branches of “pipes”) across one or more LLMs with structured I/O, validation, and clear contracts. PLX—the Pipelex language—is an open standard for declaratively specifying AI workflows.

• pipelex-cookbook — Examples & recipes Runnable pipelines you can clone and tweak: quick starts, demos, and best-practice patterns.

• pipelex-starter — New project template A template repo to bootstrap your Pipelex project with batteries included: Makefile, env setup, tests, and a minimal pipeline.

• cocode — AI-powered code analysis & docs CLI Command-line tool to analyze local or GitHub repositories, convert codebases into AI-friendly text formats, extract interfaces/imports, and automate docs & release chores—implemented as Pipelex pipelines.

# Clone and install
git clone https://github.com/Pipelex/pipelex-cookbook.git
cd pipelex-cookbook
make install   # creates a virtual environment and uses uv to install pipelex and deps

# Configure secrets
cp .env.example .env
# Add e.g. OPENAI_API_KEY=... (others optional depending on examples)

# Run the Hello World
python quick_start/hello_world.py

🔧 Tip: Install vscode-pipelex for syntax highlighting and a great editing experience in .plx files (VS Code, Cursor, Windsurf, BlackboxAI, and most forks).

1. Open pipelex-starter and click Use this template.

2. After GitHub creates your repo:

   git clone <your-new-repo-url>
   cd <your-new-repo>
   make install
   cp .env.example .env  # add your keys (OPENAI_API_KEY is enough to start)

3. Install vscode-pipelex and start building .plx pipelines under pipelines/ (or your preferred structure).

# Install the library
pip install pipelex          # or: poetry add pipelex | uv pip install pipelex

# Optional extras for providers/models
pip install "pipelex[anthropic,google,mistralai,bedrock,fal]"

Requirements:

• Python ≥ 3.10 for the core library.
• Add provider API keys to your environment (.env or CI secrets).
• Use vscode-pipelex to edit .plx files.

See the Docs for the PLX language spec, LLM integrations, and composition patterns.

# Install the CLI
pip install cocode
# Explore commands
cocode --help

Use cocode to analyze local or GitHub repositories and generate AI-friendly summaries, interfaces, and docs.

We ❤️ contributions across the ecosystem:

• Core library: see CONTRIBUTING.md in pipelex (dev setup, tests, PR process).
• Examples/recipes: add new pipelines under pipelex-cookbook (start with examples/wip/<your-folder> and include a short README in your PLX).

• 📚 Docs: https://docs.pipelex.com/
• 🐛 Issues: use the repo issue trackers (pipelex, pipelex-cookbook, pipelex-starter)
• ✉️ Security/Privacy: [email protected]

• Repos are generally MIT (see each repo’s LICENSE).
• “Pipelex” is a trademark of Evotis S.A.S.