agentleFS
michele-orsini-71michele-orsini-71/claude-skills/incremental-build/SKILL.md

incremental-build

Skill0 starsChanged 6 months agoMITView on GitHub

Adopt an incremental, vertical-slice implementation posture. Plan only at high level (which modules, rough architecture, how modules communicate), then build thin slices that each compile and run on their own, introducing types and classes only when a slice first needs them. Build one slice, run it, observe, then continue — no waterfall, no up-front contracts, no integration at the end. Optionally writes a small high-level progress plan so a later session can resume. Use when the developer moves from a brief or discussion into implementation and invokes /incremental-build, or asks to implement in baby steps, in vertical slices, or incrementally without rushing ahead.

1 other project uses a version of this.

Adopted by

Most recently taken up by

  • Other versions1 project uses a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 1

    1. microsoft/testfx★ 1.1ksame 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.

  • In this repository10 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 3 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

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

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

    After presenting the slice list, ask once whether to write a small plan file (for example PLAN.md at the project root) that tracks …

    line 30
  2. Distinctive:

    If the developer accepts, write the file with only high-level content, so it stays valid as the code changes.

    line 34
  3. Distinctive:

    Not a heavyweight planner.

    line 47
  4. Distinctive:

    Not a state machine. There is no fixed-shape report, no per-turn protocol, no filename convention beyond the single optional plan file.

    line 48
  5. Very specific:

    Not a scoping session. If scope is genuinely unclear, say so and point the developer at /brief-chat or /grill-me — do not start …

    line 49

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “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

The developer builds in baby steps: agile, not waterfall. Scope is already settled — it lives in the brief, the discussion above, or whatever the developer points you at. Do not re-open scoping and do not interview the developer. No grilling.

Adopt the following posture for the whole implementation and hold it until the developer releases you from it.

The rules

  1. High-level plan only. Show which modules you will touch, the rough architecture, and how the modules talk to each other. Do not define types, classes, interfaces, or data shapes up front. Do not create a "types" or "contracts" phase. A shape defined before the code that needs it is waterfall in miniature.
  2. Vertical slices, not horizontal layers. Lay out the work as a sequence of thin slices. Each slice cuts through every layer it needs, compiles, and runs on its own, producing something the developer can launch and observe. A slice is never "the types" or "the service layer" or "the UI" — it is "this one behaviour, end to end, minimally". Do not build the pieces separately and integrate them at the end.
  3. Shapes arrive when needed. Introduce a type, class, or interface only inside the slice whose code first needs it, and only as much of it as that slice needs. When a later slice needs more, grow the shape then. Never design the full shape in advance.
  4. One slice at a time. Build one slice, run it, see what happens, and only then decide the next. Do not rush ahead with the details of later slices. The plan can change once the first slice runs.
  5. Stop and let the developer look. After each slice, stop so the developer can launch and review it. Do not begin the next slice until the developer says to continue. If a slice turns out bigger than expected, stop and say so — do not silently absorb the extra work into the current slice.
  6. The developer owns the code. The developer may take the keyboard at any time. If the developer says "I am driving", show the code in the chat and explain what to do; do not write to files until told. If they want you to stop volunteering the next step and act as a reference desk only while they build, that is /practice-mode — invoke it and this slice rhythm still holds.

How to start

Show the high-level plan and the ordered list of slices — for each, a title and what runs at the end of it. Then offer the optional progress plan below (one short question), and stop. The developer says which slice to begin.

Optional: a high-level progress plan

After presenting the slice list, ask once whether to write a small plan file (for example PLAN.md at the project root) that tracks high-level progress, so a later session can resume without re-deriving context.

(Recommended for work that will span more than one session; skip it for a small change that finishes in one sitting. A plain "no" skips it.)

If the developer accepts, write the file with only high-level content, so it stays valid as the code changes. It is a journal, not a specification — never put types, class names, function signatures, or code shapes in it. Suggested sections:

  • Goal — what the work achieves, in prose.
  • Chosen design — the architecture at module altitude, and the alternatives rejected and why (one line each). Optional for small work.
  • Working principle — a reminder that the build is vertical slices, not horizontal layers; shapes are born in the slice that first needs them.
  • Slices — one entry per slice, each with a status marker and a one-line "verify" (what runs when it is done). Use the legend [ ] not started · [~] in progress · [x] done.
  • Decisions deferred to their slice — open choices, each tagged with the slice that will settle it and a proposed default. Do not resolve them early.
  • Current status — the last line: what is done and the single next action. This is what a resumed session reads first.

Maintenance rule: as each slice lands, flip its status marker and update the Current status line. Do not rewrite the plan or add code-level detail to it — if you find yourself editing the plan on every code change, it has become too low-level. Keep it at the altitude where the first code change cannot make it obsolete.

What this skill is not

  • Not a heavyweight planner. The slice list lives in the conversation; the only file it may produce is the opt-in high-level plan above, and only if the developer asks for it.
  • Not a state machine. There is no fixed-shape report, no per-turn protocol, no filename convention beyond the single optional plan file. The discipline is the posture, nothing more.
  • Not a scoping session. If scope is genuinely unclear, say so and point the developer at /brief-chat or /grill-me — do not start interviewing.

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.