agentleFS
librosdeialibrosdeia/builder-method-claude-plugin/skills/build/SKILL.md

build

Skill0 starsChanged 5 days agoMITView on GitHub

Phase 4 of the Builder Method — turn the PRD into the first working version of the product. Plan the tasks (each covering user flows), set up the repository, then build task by task with tests and a checkpoint commit after each one. Use when the PRD (and design) are ready and the user wants to build, or runs /builder:build.

  • Reads credentials

163 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use backprop, finances and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • JuliusBrusseeSkill

    backprop

    Bug → spec protocol. When a bug is found or a test fails, trace the cause, decide whether a new §V invariant would catch recurrence, append to §B. This is the one non-obvious thing SDD does that plan-then-execute doesn't. Triggers on test failure, bug report, post-mortem, or explicit user ask.

    Also in 4 projects that use this

  • GerardoRdz96Skill

    finances

    Use when Gera talks about his money — logging an expense ("I spent X on Y"), checking where he stands this pay-period, closing a period, planning debt payoff, onboarding his financial picture, or asking for budgeting/finance opinions. Servy's financial brain. Triggers — "I spent", "log this expense", "where am I this period", "close the period", "how's my budget", "debt plan", "should I pay off", "finances onboard", "/finances".

    Also in 3 projects that use this

  • cpiprintSkill

    favicon

    Generate favicons from a source image

    Also in 3 projects that use this

  • madeehahassanSkill

    Go-Live

    Pre-launch checklist. Use before making any app public, sharing a link externally, or announcing a product launch. Adapts to Claude Code, claude.ai, and Cowork.

    Also in 3 projects that use this

  • JuliusBrusseeSkill

    deepen

    Optional design-improvement pass for when you have spare usage to drain. Finds the shallowest modules in the code the spec touches, researches a deeper design, and proposes refactors that shrink interfaces and hide decisions — behavior held constant, tests green before and after. Proposes §I/§V/§T edits, never silent rewrites. Triggers when the user says "deepen this", "improve the design", "this module feels shallow", "pull complexity down", "use spare budget on the codebase", or invokes /ck:deepen. Leans on the codebase-design skill's deep-module vocabulary when present.

    Also in 3 projects that use this

  • synaptiaiSkill

    gate

    Create a BDSK review_gate artifact. Defines a checkpoint that controls whether generated output may proceed toward acceptance. Use when defining quality, security, or scope gates for execution plans.

    Also in 5 projects that use this

The file

Show its rules and text

We split this skill into the rules it gives an agent

13 rules. 0 are common, 6 are distinctive, and 7 are very specific to this project.

  • 0

    Common

    Most of the 149 edited versions we read keep these.

  • 6

    Distinctive

    This author's own choices. Worth a look.

  • 7

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Coverage check (mandatory): every flow id in state.flows must appear in at least one task's Covers: line.

    line 34
  2. Very specific:

    Write builder/04-plan.md (if it already exists, follow "Existing outputs" in state-rules.md: revise it, never overwrite the user's edits).

    line 37
  3. Distinctive:

    Get approval before setup: If the user asked for defaults, or already approved, continue.

    line 40
  4. Distinctive:

    Before any product code:

    line 44
  5. Distinctive:

    Git: If the folder is already a git repository (git rev-parse --is-inside-work-tree), reuse it.

    line 45
  6. Distinctive:

    Existing files: Never overwrite or delete existing source files.

    line 46
  7. Very specific:

    Check the result yourself: run the tests (front end and back end), and make sure the task's Check: holds.

    line 64
  8. Very specific:

    Commit: builder: T<n> <title>.

    line 65
  9. Very specific:

    Every decision the PRD didn't cover goes to builder/decisions.md (template ${CLAUDE_PLUGIN_ROOT}/skills/build/templates/decisions.md) as a …

    line 66
  10. Distinctive:

    Add realistic sample data, so the user can try the product straight away.

    line 68
  11. Very specific:

    If you're blocked (missing account, failing dependency, unclear requirement), stop.

    line 70
  12. Very specific:

    When every task of the current cycle is done and the app starts with the dev script:

    line 74
  13. Distinctive:

    update the progress page;

    line 77

“Common” is measured: kept by at least 6 in 10 of the 149 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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Phase 4 · Build

Goal: the first usable version of the product, working but not yet validated. Input: builder/02-prd.md (+ builder/03-design.md if there are screens). Output: the product's code, builder/04-plan.md, builder/decisions.md, and tasks tracked in state.json.

Read and follow:

  • ${CLAUDE_PLUGIN_ROOT}/skills/_shared/tone.md
  • ${CLAUDE_PLUGIN_ROOT}/skills/_shared/state-rules.md
  • ${CLAUDE_PLUGIN_ROOT}/skills/_shared/gates.md
  • ${CLAUDE_PLUGIN_ROOT}/skills/build/references/prompts.md (the method's original intent)

0 · Can we build here?

This phase runs commands (installing, testing, git). If you don't have a shell tool (no Bash), don't try to work around it. Say this, in the user's language, and stop:

Building needs a terminal. Open this same folder in Claude Code and run /builder:next. Everything done so far is saved in builder/, so you'll continue exactly where you are.

1 · Gate

Apply the Build gate from gates.md. This includes the hasFrontend: null question, and generating the missing PRD or design with Continue anyway. A PRD generated here must have numbered flows (### F<n> —) with acceptance criteria, and they must be mirrored into state.flows, so the plan can cover them.

Then go to the first step of this phase that isn't done (plan → setup → build). Set phases.build.status to in_progress and currentPhase to build.

Step 1 · plan

Delegate to the planner subagent (Agent tool, subagent_type: "builder:planner"). Give it the paths of the PRD, the design (if any) and state.json. It returns the plan in the format of ${CLAUDE_PLUGIN_ROOT}/skills/build/templates/04-plan.md. If the subagent tool isn't available, write the plan yourself, following ${CLAUDE_PLUGIN_ROOT}/agents/planner.md.

Coverage check (mandatory): every flow id in state.flows must appear in at least one task's Covers: line. If one is missing, add or adjust a task before going on. Never show an uncovered plan as finished.

Then:

  1. Write builder/04-plan.md (if it already exists, follow "Existing outputs" in state-rules.md: revise it, never overwrite the user's edits).
  2. Fill state.tasks ({ id, title, flows, status: "pending", cycle }), and each flow's tasks list.
  3. Explain the plan in a few plain lines: the stack and why, how many tasks there are, and what the user will be able to try first.
  4. Get approval before setup. If the user asked for defaults, or already approved, continue.

Step 2 · setup

Before any product code:

  • Git. If the folder is already a git repository (git rev-parse --is-inside-work-tree), reuse it. Never run git init again. Otherwise run git init, and explain in two lines what a commit is (a saved point you can go back to).
  • Existing files. Never overwrite or delete existing source files. Build around them, or ask.
  • Project files. Create:
    • the directory structure;
    • CLAUDE.md, from ${CLAUDE_PLUGIN_ROOT}/skills/build/references/claude-md-base.md, filled in (merge with any existing CLAUDE.md, never replace it);
    • the dev, test and build scripts;
    • linting and formatting;
    • .gitignore, covering at least .env, .env.* (but not .env.example), dependencies and build output;
    • .env.example, if the product needs secrets.
  • Everything follows the plan and the PRD's validation criteria. Prefer boring, well-known tools.
  • First checkpoint commit: builder: setup.

Step 3 · build

Work through the tasks of the current cycle in plan order. Keep going autonomously until all are done and tested.

For each task:

  1. Set it to in_progress in state.json.
  2. Delegate it to the implementer subagent (Agent tool, subagent_type: "builder:implementer", with the model the plan names). Give it the task text, the flows it covers with their acceptance criteria, and the relevant design screens. If the subagent tool isn't available, do the task yourself, following ${CLAUDE_PLUGIN_ROOT}/agents/implementer.md.
  3. Check the result yourself: run the tests (front end and back end), and make sure the task's Check: holds.
  4. Commit: builder: T<n> <title>. Record the short hash in the task's checkpoint, set it to done, and update the progress page.
  5. Every decision the PRD didn't cover goes to builder/decisions.md (template ${CLAUDE_PLUGIN_ROOT}/skills/build/templates/decisions.md) as a new D<n> with Status: pending. Increase state.decisions.pending and state.decisions.total by one each time.

Add realistic sample data, so the user can try the product straight away.

If you're blocked (missing account, failing dependency, unclear requirement), stop. Explain the blocker in plain words, say what you need from the user, and leave the task in_progress.

Finish

When every task of the current cycle is done and the app starts with the dev script:

  • mark build and its steps done;
  • set currentPhase to iterate;
  • update the progress page;
  • tell the user how to open the app, and that the next phase checks it together, flow by flow.

Close by suggesting /builder:next.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.