15:I[981,["658","static/chunks/658-a698693f56d0116a.js","183","static/chunks/183-6beeeb801c274aaa.js","474","static/chunks/app/%5Btype%5D/layout-631c4730b83c8223.js"],"Posts"] 16:T7b4,I am working on a lightweight agentic workflow to accelerate software development on my own machine. I am executing AI-generated code locally on my machine in a sandboxed environment, and feeding the errors back into the LLM automatically.

Has anyone done anything similar on their own machine? I am interested to hear other thoughts.

General overview of the current workflow (without a lot of the finer details):

Call Gemini API to generate Python code for specific function/problem ->

Run AI-generated code in Docker container ->

Take any runtime/compilation errors and feed back to Gemini ->

Run hardcoded tests for functionality and send results back to Gemini ->

Repeat step 1 until max iterations are met (or testing passes)

I had a few general questions:

1) What patterns, antipatterns, or architectures do people find best for these workflows?

2) Is using Docker considered the easiest and safest way of quarantining AI-generated code?

3) I have read a couple posts online about “compressing” the context of previous changes through quick summaries. We can also use vectors for traversing documentation to speed up context retrieval for the AI. Does anyone have any general advice here on what works and what doesn’t?

4) Should I have an external Agent that reviews the errors of the previous iterations to see if Gemini is falling into a loop? Sometimes when I use LLMs for coding, I notice that they fall into loops (they will just alternate between two buggy solutions). Should we just break the iterations in this case and rely on human intervention?

I am also trying to utilize AutoML packages (like FLAML) with this workflow. I am implementing this workflow to perform “automatic” data analysis on datasets to get better predictions. Obviously, I understand that this will not perform as well

Digital Threat Modeling Under Authoritarianism

How England misplaced its first king

Avoiding a Surprise AWS Bill

There's no such thing as plain text [video]

I built thumbnai.com for a friend, he told me to double down on it

Why, in old books, are dates often given with the years redacted? (2012)

Original Startup Garage: Hobbyists, Hackers, and Home PCs

Digital ID, same place, different time

A primer for using weather surveillance radar to study bird migration

OpenPrinter, an Open Hardware Printer

The Moon Landing: An Undelivered Nixon Speech

F-Droid and Google's Developer Registration Decree

Ask HN: Best Architecture Patterns for Lightweight SWE Workflows?

Bad Apple but It's Played Inside Super Mario Bros. [TAS]

An AI Index for all our customers

Off-duty DC firefighter was shot and dialed 911, but no one picked up

Stealing from Google

6,100-Qubit Processor Shatters Quantum Computing Record

AI Platform Citation Patterns: What Marketers Must Know About GenAI Discovery

Reddit Seeks to Strike Next AI Content Pact with Google, OpenAI

Forecasting Revolutionnay Package

The Internet Is Powered by Generosity

Startup Nights 2025 is comming up on 6-7 Nov. in Switzerland

Twenty Lessons on Fighting Tyranny from the Twentieth Century

Building a Custom Car Computer in 2006: Honda Accord PC Integration Project

Germany: Ministry of Economic Affairs effectively abolishes supply chain law

Animating Geometry with AMD DGF

Nvidia Vulkan Ray Tracing Tutorial

Figma Rendering: Powered by WebGPU

Show HN: When you open Instagram, your textbook opens instead – ZenScript App

Show HN: Traceroute Visualizer