Intel / article

I rebuilt a 4-year-old app in 5 days using many agents – here's harness

Source: mega.dev
Author: gregrog
Date: Published Sep 14, 2026

Context

The author of this five-day app rebuild account describes substantial preparation around the coding agents. Shared project specifications, a task board and unified logging gave the agents persistent , and a testing setup connected to the app's interface let them take screenshots and measurements. Researcher agents gathered context and coordinator agents drafted specifications and divided work into waves, while the author kept the project vision and supplied architectural feedback. The first-person account explains how these pieces supported the rebuild.

The transferable idea is to make project knowledge and feedback available where agents work. The reported delivery time belongs to this particular rebuild, backed by the author's own setup and oversight. The author describes the work as requiring a great deal of effort and expertise and says product development is not yet fully autonomous.

Terms in this piece · Glossary

context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
multi-agent — Using several AI agents on one problem — splitting work in parallel, checking each other, or filling different roles like planner and reviewer.

Read the source mega.dev