--- title: How I turned fstack into a software factory | daily.dev description: Flavio Copes describes how he evolved his fstack agent skill system into a multi-agent software factory. The approach uses GPT-5.6 Sol as a... canonical: https://daily.dev/posts/how-i-turned-fstack-into-a-software-factory-1ssqkyumx twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: How I turned fstack into a software factory | daily.dev og:description: Flavio Copes describes how he evolved his fstack agent skill system into a multi-agent software factory. The approach uses GPT-5.6 Sol as a... og:url: https://daily.dev/posts/how-i-turned-fstack-into-a-software-factory-1ssqkyumx og:image: https://api.daily.dev/og/posts/1ssQkyUmX.png og:image:alt: How I turned fstack into a software factory og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # How I turned fstack into a software factory **[Flavio Copes](https://daily.dev/sources/flaviocopes)** · 21 min read · 1 upvotes · 0 comments ## Summary Flavio Copes describes how he evolved his fstack agent skill system into a multi-agent software factory. The approach uses GPT-5.6 Sol as a manager/coordinator, Composer subagents as bounded production cells, and a structured pipeline: idea validation, product context capture in AGENTS.md, planning in PLAN.md, parallel execution with strict file ownership, and mandatory review gates (ACCEPT/PATCH/RE-BRIEF). Key principles include keeping long-lived state in the repository rather than chat, preventing workers from making architectural decisions, using a dependency graph before parallelizing, and maintaining a simplification station to counteract code bloat. The system standardizes the production process without homogenizing the products, keeping human judgment at the center of decisions about what to build and whether results are acceptable. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### How should I split work between a manager model and worker subagents when building software with multiple AI agents? Use one strong reasoning model as the planner, coordinator, and reviewer, and only delegate bounded, independently verifiable tasks with clear inputs, constraints, and output format to worker subagents. Workers must not redefine the plan, change architecture, expand scope, or touch shared contracts unless explicitly instructed; they should flag uncertainty rather than guess, and every return should be reviewed as ACCEPT, PATCH, or RE-BRIEF before integration. _Developers designing multi-agent coding workflows can compare approaches like this on daily.dev before committing to a setup._ ### What should go into a brief given to an AI coding subagent to avoid vague or incorrect implementations? A self-contained brief needs five parts: a one-sentence goal, context the worker cannot safely infer, exact deliverables and files to change, explicit constraints on what not to touch, and a required 'bring back' report covering files inspected, files changed, tests run, decisions made, and uncertainties. The brief should end by instructing the worker to flag uncertainty rather than guess silently. _Teams refining prompts for coding agents can track brief-writing patterns like this on daily.dev._ ### What are the three checks an AI-driven code review gate should run before accepting a change? A quality gate should ask whether the change actually works (proven by running tests, builds, or browser checks rather than just claiming success), whether it matches the approved plan (checking for missing or unapproved extra work and blast radius), and whether it is simple (flagging unnecessary abstractions, unused code, or dependencies without fixing them automatically, keeping review and repair separate). _Engineers building review gates for agent-generated code can compare quality-gate designs on daily.dev._ ## Similar posts on daily.dev - [Superpowers, GSD, and GSTACK: Picking the Right Framework for Your Coding Agent](https://daily.dev/posts/superpowers-gsd-and-gstack-picking-the-right-framework-for-your-coding-agent-jn14rxumh) · Pulumi · 1 upvotes · 0 comments - [Three Years of Building Agents in Production \(Part 1\)](https://daily.dev/posts/three-years-of-building-agents-in-production-part-1--d2wjasjzx) · Mabl Engineering Blog · 3 upvotes · 1 comments - [Trying the Software factory pattern.](https://daily.dev/posts/trying-the-software-factory-pattern--3wqdkev6u) · Irrational Exuberance · 3 upvotes · 0 comments --- Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#gpt](https://daily.dev/tags/gpt) [View this post on daily.dev](https://daily.dev/posts/how-i-turned-fstack-into-a-software-factory-1ssqkyumx) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"How I turned fstack into a software factory","url":"https://daily.dev/posts/how-i-turned-fstack-into-a-software-factory-1ssqkyumx","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/how-i-turned-fstack-into-a-software-factory-1ssqkyumx"},"datePublished":"2026-07-29T12:44:01.213Z","dateModified":"2026-09-14T07:03:13.151Z","description":"Flavio Copes describes how he evolved his fstack agent skill system into a multi-agent software factory. The approach uses GPT-5.6 Sol as a...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/7219243bb6b6113ecc90bd933965a4ef?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/7219243bb6b6113ecc90bd933965a4ef?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Flavio Copes","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Flavio Copes","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/1f2dbd275797411bbd37bfdaaf0951a9","url":"https://daily.dev/sources/flaviocopes"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/how-i-turned-fstack-into-a-software-factory-1ssqkyumx","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai-agents,gpt","timeRequired":"PT21M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Flavio Copes","item":"https://daily.dev/sources/flaviocopes"},{"@type":"ListItem","position":3,"name":"How I turned fstack into a software factory"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/how-i-turned-fstack-into-a-software-factory-1ssqkyumx#faq","mainEntity":[{"@type":"Question","name":"How should I split work between a manager model and worker subagents when building software with multiple AI agents?","acceptedAnswer":{"@type":"Answer","text":"Use one strong reasoning model as the planner, coordinator, and reviewer, and only delegate bounded, independently verifiable tasks with clear inputs, constraints, and output format to worker subagents. Workers must not redefine the plan, change architecture, expand scope, or touch shared contracts unless explicitly instructed; they should flag uncertainty rather than guess, and every return should be reviewed as ACCEPT, PATCH, or RE-BRIEF before integration. Developers designing multi-agent coding workflows can compare approaches like this on daily.dev before committing to a setup."}},{"@type":"Question","name":"What should go into a brief given to an AI coding subagent to avoid vague or incorrect implementations?","acceptedAnswer":{"@type":"Answer","text":"A self-contained brief needs five parts: a one-sentence goal, context the worker cannot safely infer, exact deliverables and files to change, explicit constraints on what not to touch, and a required 'bring back' report covering files inspected, files changed, tests run, decisions made, and uncertainties. The brief should end by instructing the worker to flag uncertainty rather than guess silently. Teams refining prompts for coding agents can track brief-writing patterns like this on daily.dev."}},{"@type":"Question","name":"What are the three checks an AI-driven code review gate should run before accepting a change?","acceptedAnswer":{"@type":"Answer","text":"A quality gate should ask whether the change actually works (proven by running tests, builds, or browser checks rather than just claiming success), whether it matches the approved plan (checking for missing or unapproved extra work and blast radius), and whether it is simple (flagging unnecessary abstractions, unused code, or dependencies without fixing them automatically, keeping review and repair separate). Engineers building review gates for agent-generated code can compare quality-gate designs on daily.dev."}}]} ```