{"id":51872708,"url":"https://github.com/rohansx/build-discipline","last_synced_at":"2026-07-25T01:30:28.123Z","repository":{"id":369067361,"uuid":"1288216014","full_name":"rohansx/build-discipline","owner":"rohansx","description":"A language and domain-agnostic engineering discipline you drop into your AI coding flow so it builds like a senior engineer instead of vibe-coding a demo that breaks.","archived":false,"fork":false,"pushed_at":"2026-07-03T11:50:17.000Z","size":10,"stargazers_count":1,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"master","last_synced_at":"2026-07-03T13:27:37.578Z","etag":null,"topics":["agents","claude","claude-code","claude-skills","code","cursor","llm","skills","software-engineering"],"latest_commit_sha":null,"homepage":"","language":null,"has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/rohansx.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-07-03T11:34:45.000Z","updated_at":"2026-07-03T11:57:04.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/rohansx/build-discipline","commit_stats":null,"previous_names":["rohansx/build-discipline"],"tags_count":null,"template":false,"template_full_name":null,"purl":"pkg:github/rohansx/build-discipline","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/rohansx%2Fbuild-discipline","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/rohansx%2Fbuild-discipline/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/rohansx%2Fbuild-discipline/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/rohansx%2Fbuild-discipline/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/rohansx","download_url":"https://codeload.github.com/rohansx/build-discipline/tar.gz/refs/heads/master","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/rohansx%2Fbuild-discipline/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":35862474,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-07-20T02:08:10.276Z","status":"online","status_checked_at":"2026-07-24T02:00:07.870Z","response_time":62,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["agents","claude","claude-code","claude-skills","code","cursor","llm","skills","software-engineering"],"created_at":"2026-07-25T01:30:22.255Z","updated_at":"2026-07-25T01:30:28.084Z","avatar_url":"https://github.com/rohansx.png","language":null,"funding_links":[],"categories":[],"sub_categories":[],"readme":"# build-discipline\n\nA single, language- and domain-agnostic engineering discipline you drop into your\nAI coding flow (Claude Code, Cursor, or any agent that reads skills / rules files)\nso it builds like a senior engineer instead of vibe-coding a demo that breaks on\ncontact with reality.\n\nIt's one file: [`SKILL.md`](./SKILL.md). That's the whole thing.\n\n## What it's for\n\nModels are good at producing plausible-looking code fast. What they skip by\ndefault is the *discipline* that separates a demo from something that survives\nreal use: measuring the constants they'd otherwise guess, separating state so it\ncan't go stale, checking correctness with invariants instead of hope, and\nassembling in gated stages. This skill externalizes that discipline into the\ncontext so it happens by default.\n\nIt doesn't make a model \"smarter.\" It makes the model *do the things a careful\nengineer does* — which is a more honest and more useful claim.\n\n## What's inside\n\nNine load-bearing principles, each earned from a real failure, each concrete\nenough to apply to code you're writing right now — regardless of language\n(Rust, TS, Python, Go…) or domain (services, parsers, pipelines, simulators,\ntools, agents):\n\n1. Separate source-of-truth / derived / ephemeral state — never persist what you\n   can derive.\n2. Keep a pure, seeded, reproducible core so you can actually test it.\n3. **Calibrate every gating constant from printed data, never from intuition.**\n   (The one that pays for the whole file.)\n4. Prefer composable, data-driven rules over special-cases and global reads.\n5. Assert invariants, not outputs — build the harness before the UI.\n6. Isolate → stabilize → compose → tune → verify, gating every step.\n7. Every automated judgment carries its reason; destruction needs a guard.\n8. Coupled constants can be jointly impossible — check interacting ranges.\n9. Design the degenerate case; make even the worst state coherent.\n\nPlus a short section on applying it turn-by-turn inside an AI-assisted build loop.\n\n## How to use it\n\n**As a Claude Agent Skill:** install `SKILL.md` as a skill (or upload the packaged\n`.skill` file) and it triggers automatically on real coding tasks.\n\n**In any coding agent:** point your project rules at it — e.g. reference or paste\n`SKILL.md` into `CLAUDE.md`, `.cursorrules`, or your agent's system context. It's\nplain Markdown with nothing tool-specific.\n\n**As a checklist for humans:** read it before starting a non-trivial build, and\nkeep the nine principles open while reviewing a PR.\n\n## Provenance\n\nDistilled from two real from-scratch builds — a self-improving knowledge system\nand a multi-agent emergence simulator — that hit *the same underlying mistakes*\nfrom opposite domains. The failure modes named in the skill (a threshold that\nsilently produced zero output, a drain/refill mismatch hidden behind healthy\naverages, a pair of radii that made an interaction geometrically impossible) all\nactually happened and were actually fixed. The specifics are the value.\n\n## License\n\nMIT.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Frohansx%2Fbuild-discipline","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Frohansx%2Fbuild-discipline","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Frohansx%2Fbuild-discipline/lists"}