Written policies plus a 'Grumpy Engineer' review agent for qualitative code judgments linters can't express
Some rules can't be linted — a three-state machine ballooned to ten, code too defensive, tests too numerous but weak. Write these as prose policies for the repository, then run a review skill (internally called 'Grumpy Engineer', 'Garfield') that spins up a sub-agent per policy; each sub-agent either gives feedback or stands down, looping until every one stands down, after which a human reviews.
online B2B Service / Consulting / Agency
From AI Engineer — Stop Prompting — Greg Pstrucha, Sentry at 12:22
Problem: Qualitative taste — defensiveness, state-machine bloat, test quality — cannot be captured by deterministic linters, yet agents violate it endlessly.
For: Engineering teams whose agents pass lint and tests but produce qualitatively bad code.
Products from this video
Examples
- Sentry repository policies: 'I don't want too many tests, I want stronger tests' and 'I don't want code that's too defensive', enforced by the Grumpy Engineer review loop.
Other takes on AI implementation services
- AI voice-agent consulting service for local small businesses
- Enterprise document question-answering system grounded in company policies
- AI implementation and automation consulting for established businesses
- Managed AI agent service for small businesses
- AI-powered missed-call text-back service for small businesses
- Missed-call text-back service for roofing companies