Pipelex

Declarative language for composable Al workflows. Devtool for agents and mere humans.

[Louis Choquel @lchoquel]

@garrytan Go beyond skills: with MTHDS you get deterministic orchestration of multiple AI jobs + structured generation and composability. It's the missing layer between Skills and Code. It installs into Claude Code with a simple copy-paste, just like gstack --> https://t.co/XzrVEUPApv

[Louis Choquel @lchoquel]

4 months of heads-down building. Biggest Pipelex release yet.

New open language spec called MTHDS. A Claude Code plugin that writes, tests, and visualizes AI workflows for you. One API key for all models. A package manager to share it all.

Full drop 🧵

[Louis Choquel @lchoquel]

Without Know-How, your agent is a zombie 🧟‍♂️.
A Knowledge Graph isn't enough.
→ We need a Know-How Graph ‼️

⏩ A network of executable methods that AI can actually reuse.
Read our Viewpoint 👇 ➡️ knowhowgraph (dot) com

[Dev Hunt @devhunt_]

🎉🏆 Winners of the last week's DevHunt! 🎉

Our top 3 champs are:

1⃣🥇 @PipelexAI – best Dev Tool of the week!!!
▶️ https://t.co/CCmbRg8c2N

2⃣ @kixxauth 🥈
▶️ https://t.co/3SnbKQQqqW

3⃣ Pulsetracker by @laggwalid 🥉
▶️ https://t.co/H1htEIP8AI

Keep up the great work, everyone! 👍

[Founder Journeys @FounderJourneys]

@PipelexAI Listen to @lchoquel talk about how Pipelex will change building of AI workflows forever, on Founder Journeys Pod on Spotify for Creators https://t.co/fWEo3BzErN

[Founder Journeys @FounderJourneys]

New Episode! Presenting Pipelex, The open standard language simplifying AI workflows for everyone! https://t.co/J6if7gqzeb @lchoquel @PipelexAI

[- @Danara_AI]

9pm Wednesday at @PipelexAI evening hackathon

I was building invoice OCR recognition tool

Among demos - Zillow listing scam detector, calorie counter and Free Food events finder in #SanFrancisco

[Louis Choquel @lchoquel]

First Pipelex hackathon in SF last night ⚡

50+ builders. 4.5 hours. Dozens of AI workflows built from scratch.

The energy was unreal 🧵

[Louis Choquel @lchoquel]

AI that builds AI 🌀

We can vibe-code entire apps in 4 prompts, but still spend days writing custom code for one AI workflow that barely hits 80% reliability.

That's backwards. So we fixed it.

[Louis Choquel @lchoquel]

AI doesn't just need agents - it needs workflows.

🚀 Today we're launching Pipelex: an open-source DSL + Python runtime for repeatable AI workflows.

Think Dockerfile/SQL for LLM pipelines - declare steps, swap models, let the runtime handle execution.

[Thomas Hebrard @thomashebrard]

🚀 Hey builder! Glad to tell you that today we're launching @PipelexAI : an open-source DSL + Python runtime for repeatable AI workflows.

AI doesn't just need agents - it needs workflows.
Most knowledge work follows patterns. If you know the path, you don't need an expedition… you need a bullet train.

Think Dockerfile/SQL for LLM pipelines - declare steps, swap models, let the runtime handle execution.

What makes it different:
Declarative, not glue code
Agent-first: LLMs can run, audit & build workflows (via MCP)
Open & composable: MIT license, works with any provider

What's included:
→ Python runtime
→ FastAPI server + Docker
→ MCP server (agent integration)
→ n8n node
→ VS Code extension

I would love your feedback, don't hesitate to test it and tell me.
(links in bio)

[Thomas Hebrard @thomashebrard]

[LAUNCH WEEK - day #1]
AI doesn't just need agents - it needs workflows.
Most knowledge work follows patterns. If you know the path, you don't need an expedition… you need a bullet train.

🚀 Today we're launching @PipelexAI : an open-source DSL + Python runtime for repeatable AI workflows.

Think Dockerfile/SQL for LLM pipelines - declare steps, swap models, let the runtime handle execution.

What makes it different:

Declarative, not glue code
Agent-first: LLMs can run, audit & build workflows (via MCP)
Open & composable: MIT license, works with any provider

What's included:
→ Python runtime
→ FastAPI server + Docker
→ MCP server (agent integration)
→ n8n node
→ VS Code extension

Try it, break it, tell us what hurts. Feedback = gold.
(links in bio)

[Dev Hunt @devhunt_]

→ @PipelexAI is a declarative language and Python runtime for repeatable, agent-first AI workflows. It defines LLM pipelines like Dockerfile or SQL — describe steps and interfaces, and any model or provider can execute them.

🔴 Live on Dev Hunt → https://t.co/t2SVRAghDp

[Pipelex @PipelexAI]

Pipelex hasn't launched any cryptocoin

If you see a Pipelex AI cryptocoin, it's a scam

[Louis Choquel @lchoquel]

AI Isn't Real Intelligence? Bullshit.

[Tom Dörr @tom_doerr]

open-source tool for building repeatable AI workflows with modular, reusable components

[Louis Choquel @lchoquel]

Try Cocode, the friend of your code https://t.co/uIPYwPZ7QZ

[Louis Choquel @lchoquel]

From whiteboard to AI workflow in less than 5 min with no hands 🙌🏻

[Python Hub @PythonHub]

Kajson

Universal JSON Encoder/Decoder for Python.

https://t.co/hekIeqlFbK

[Thomas Hebrard @thomashebrard]

Wanna reach boss level in AI workflows? Check out the @PipelexAI demo.
Pipelex is an open-source AI dev tool that helps you write the most resilient and powerful LLM pipelines using a simple language. https://t.co/AoJ88LRoEu