18:T74a,As a beginner learning to build ML models, I found it annoying to have to keep printing tensor shapes every other line, having to step through the debugger to check where did I mess up the shapes again.

So I built Trickle, it takes the data that flows through your code, caches the types and display them inline (as if you have type annotations).

The idea is: "Let types trickle from runtime into your IDE". You get types in Python without having the write them manually.

It works by rewriting your Python AST at import time — after every variable assignment, it inserts a lightweight call that records the type and value. No decorators, no code changes. Just run your script through trickle run python train.py and every variable gets its type visible.

One cool feature is Error snapshots, by toggling it in VSCode status bar, you can see the exact data that is in each variable when the error happened.

For AI agents, trickle-cli outputs the inline runtime types together with the source code so agent can better diagnose the issue.

For Jupyter notebooks: %load_ext trickle as your first cell, and every cell after is traced.

Quick try: pip install trickle-observe npm install -g trickle-cli code --install-extension yiheinchai.trickle-vscode

   trickle run python demo/demo.py
   trickle hints demo/demo.py

Limitations: - Expect 2-5x slowdown — designed for development, not production.

Also supports JavaScript/TypeScript (Express, Fastify, Koa, Hono) though the Python/ML side is where I've focused most effort.

In the future, I imagine there to be potential for using this as runtime observability in production via probabilistic sampling of types. Now, we know the code, we know the data, which is all the information we need to debug.

Happy to answer questions19:T716,I save all my chat transcripts because I believe this data will matter as a work credential (or resume substitute). Y Combinator already asks founders to submit AI co

Keep Hyper-Scale Datacenters Out of the Desert (March 2026)

Show HN: AI agent deploys an edge AI model on a microcontroller via MCP

Closing Arguments Begin in Twitter Trial Accusing Musk of Driving Down Stock

The Context Must Flow – Permit MCP Gateway Explainer [video]

The Ugliest Airplane: An Appreciation

Node.js worker threads are problematic, but they work great for us

Show HN: BulkHead – iOS File Manager for SFTP, SMB, WebDAV, and S3

Tokens Are Not Securities

Intel enables Precompiled Shader Delivery in new driver

Comprehension Debt

Three More Jury Notes in Social Media Trial Focus on YouTube Witness, Father

The problem is not generating code. It is trusting the change

Show HN: Trickle – See PyTorch tensor shapes inline in VSCode as you code

Xiaomi MiMo-V2-Omni: See, Hear, Act in the Agentic Era

GitHub and Claude Are Down Three Out of Four Days

Recovery of the first fiber-optic submarine cable across an ocean – TAT-8

How to write yourself every day

JupyterLab 4 Extensions

What's Behind 'Alpine Divorce'?

Show HN: Motif - Analyze your Cursor and Claude Code chat history

Agentic review of Linux Kernel code changes

Show HN: The-telegram-telegram (thermal printer style)

The Benefits of Conforming

Show HN: Packet28 – Context Layer for AI coding agents

Ask HN: How do you distinguish prompt issues from model failures in LLMs today

Lifeguard: A fast Rust-based static analysis tool, for Python lazy imports

AWS spurs Catch-22, ending PostgreSQL 13 support for RDS

EU Inc.: incorporate in any EU country in 48 hours with no minimum capital

Population Around a Point

Xiaomi MiMo-V2-Pro: Flagship Foundation Model Towards Agent Era

Show HN: A 30-minute course to get up to speed on how AI agents work

Comments