build-project
Skill0 starsChanged 3 months agoMITView on GitHub
Build or take over an application, website, or software project from a user idea through authoritative research, durable project records, implementation, testing, and local verification. Use only when explicitly invoked with $build-project for end-to-end project work; do not use for small isolated edits.
4 other projects use a version of this.
Adopted by
Most recently taken up by
- rohitg00/ai-engineering-from-scratchtheir own version, first seen Nov 3, 2026
- fancyboi999/ai-engineering-from-scratch-zhtheir own version, first seen Oct 25, 2026
- Shri-Phnx/claude-skillstheir own version, first seen Oct 13, 2026
- seqra/opentainttheir own version, first seen Sep 4, 2026
Other versions4 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 4
- rohitg00/ai-engineering-from-scratch★ 78ksame name
- fancyboi999/ai-engineering-from-scratch-zh★ 1.3ksame name
- seqra/opentaint★ 172same name
- Shri-Phnx/claude-skills★ 2same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
The file
We split this skill into the rules it gives an agent
10 rules. 0 are common, 10 are distinctive, and 0 are very specific to this project.
0
Common
Most of the 3 edited versions we read keep these.
10
Distinctive
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 13
Before intake, read intake and autonomy.
- Distinctive: line 14
When creating or changing project guidance, architecture, design, or code organization, read project records and design.
- Distinctive: line 15
Before research, dependency work, testing, verification, or final reporting, read quality and safety.
- Distinctive: line 21
Do not start material implementation until the user answers required human decisions or explicitly delegates eligible technical choices …
- Distinctive: line 22
After intake, proceed autonomously through local research, implementation, tests, builds, and local browser verification.
- Distinctive: line 23
Ask immediately when a newly discovered decision blocks correct work.
- Distinctive: line 24
Do not deploy, publish, spend money, create external accounts, send communications, consume paid or limited external verification, or use …
- Distinctive: line 25
Do not report completion while required checks fail, required records are stale, or mandatory user decisions remain unresolved.
- Distinctive: line 31
Keep user-facing prose direct and concrete.
- Distinctive: line 41
Done: all acceptance criteria and required gates pass locally.
“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Build Project
Deliver a coherent, locally verified project without making product-defining decisions for the user. Use the repository itself as evidence and preserve its working conventions unless the user asks to change them.
Read the right reference
- Before intake, read intake and autonomy.
- When creating or changing project guidance, architecture, design, or code organization, read project records and design.
- Before research, dependency work, testing, verification, or final reporting, read quality and safety.
Operating contract
- Inspect the repository before asking questions. Reuse facts that are already clear from source, configuration, tests, and current documentation.
- Present all presently knowable questions in a numbered initial intake. State the total number, ask one decision per question in plain language, explain the recommended option, and show how many remain. Offer:
Answer all non-human decisions automatically. - Do not start material implementation until the user answers required human decisions or explicitly delegates eligible technical choices through auto mode.
- After intake, proceed autonomously through local research, implementation, tests, builds, and local browser verification. Do not repeatedly ask for ordinary engineering choices.
- Ask immediately when a newly discovered decision blocks correct work. Defer non-blocking decisions to the finalization batch.
- Do not deploy, publish, spend money, create external accounts, send communications, consume paid or limited external verification, or use production credentials or data without explicit approval.
- Do not report completion while required checks fail, required records are stale, or mandatory user decisions remain unresolved.
Build behavior
- Establish measurable acceptance criteria before implementing stable behavior. Write tests before or alongside the implementation they specify. Do not rewrite correct tests merely to accept faulty code.
- Research current technical facts from primary sources. Record links, retrieved dates, versions, and decisions instead of downloading entire third-party manuals.
- Keep user-facing prose direct and concrete. Do not use em dashes unless the user explicitly asks for them. Avoid generic marketing filler and unsupported claims.
- For interfaces, translate the approved direction into tokens, reusable components, and named variants. Do not copy a living designer's style or solve a local need with an unexplained one-off override.
- Organize code by stable responsibilities and domain boundaries. Keep shared abstractions small and intentional; do not create a shared layer merely because two files look similar.
Continuity and finalization
Create or update the project records described in the project-records reference. Treat AGENTS.md as a stable map to current records, and keep CLAUDE.md as a one-line pointer to it. When a decision or architecture changes, update the current record, mark the old decision superseded, audit for stale instructions, and re-run affected checks.
At the end, summarize delivered behavior, verification evidence, remaining limitations, and any final decisions. Use one of these outcomes exactly in substance:
- Done: all acceptance criteria and required gates pass locally.
- Implemented, but mandatory decisions remain: explain the minimum questions required to finalize safely.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

