Mine your own agent transcripts and PR reviews into a set of custom lint rules that enforce your conventions deterministically
Agents have no memory, so after context compaction they repeat the same elementary mistakes — overly defensive code, useless tests, unnecessary endpoints — and re-prompting the correction only lasts five minutes. Instead, turn every repeated correction into a deterministic check the repo enforces on its own.
online BOTH Product / E-commerce
From AI Engineer — Stop Prompting — Greg Pstrucha, Sentry
Problem: Re-prompting corrections fails because agents forget; the same mistakes return after every compaction. Codifying rules into lint makes the repo remember instead of the agent.
For: Software engineers and teams working with AI coding agents on a real codebase.
Products from this video
Examples
- Sentry, an older Django/DRF codebase without strict typing everywhere, added lint rules so API endpoints stay in sync with the OpenAPI schema.
Other takes on Digital products
- An online sticker-making education business built around low-ticket digital products, paid bootcamps, a membership, upsells, and a physical sticker-making bundle
- A niche SaaS product that solves a personally experienced problem for a specific technical user group
- AI-powered clipping, captioning, scheduling, and social-media management SaaS for creators and marketers
- Build and monetize a simple niche calculator website
- Online trade school funded through government grants
- Build and sell a narrowly scoped AI-generated software product