agentleFS
dvy1987dvy1987/agent-loom/.agents/skills/incremental-implementation/SKILL.md

incremental-implementation

Skill4 starsChanged 3 months agoMITView on GitHub

Deliver multi-file changes in thin vertical slices — implement, test, verify, commit, repeat. Load when implementing a feature from a plan, building across more than one file, refactoring, or when tempted to land a large change in one pass. Also triggers on "incremental implementation", "vertical slice", "thin slice", "one slice at a time", "don't do it all at once". Routes single-file fixes to direct execution. Pairs with test-driven-development and git-workflow-and-versioning.

79 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use constraint-driven-development, deprecation-and-migration and 8 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.
  • MrCipherSmithSkill

    constraint-driven-development

    Use when a project needs a persistent, measurable quality bar — starting a project without documented standards, 'set up constraints', or when coverage/perf/a11y thresholds get re-debated per-PR. Produces CONSTRAINTS.md with enforceable numbers, ratchets, and a guard against silent weakening. NOT for throwaway prototypes or a one-off CI setup.

    Also in 14 projects that use this

  • HK-hubSkill

    deprecation-and-migration

    Manages deprecation and migration. Use when removing old systems, APIs, or features. Use when migrating users from one implementation to another. Use when deciding whether to maintain or sunset existing code.

    Also in 36 projects that use this

  • HK-hubSkill

    observability-and-instrumentation

    Instruments code so production behavior is visible and diagnosable. Use when adding logging, metrics, tracing, or alerting. Use when shipping any feature that runs in production and you need evidence it works. Use when production issues are reported but you can't tell what happened from the available data.

    Also in 29 projects that use this

  • seikaikyoSkill

    doubt-driven-development

    對非平凡決策啟動新鮮 context 對抗審查(找碴不背書),在修正還便宜的時候抓出錯誤方向。適用:高風險改動(production、資安敏感邏輯、不可逆操作)、不熟的程式碼、要宣稱「這樣是安全的 / 可行的」之前。機械性操作與一行修改不適用。

    Also in 33 projects that use this

  • seikaikyoSkill

    interview-me

    需求不明時的意圖萃取訪談:一次一題、每題附上自己的猜測、聽出「真正想要 vs 覺得應該要」,直到能預測使用者反應(約 95% 信心)才動工。適用:需求缺少對象 / 動機 / 成功標準 / 約束,或使用者點名「訪談我」「先確認一下」「我們確定嗎」。明確自足的指示、純資訊查詢、機械性操作不適用。

    Also in 30 projects that use this

  • thunderkdsSkill

    review-diff

    Generate a Markdown review file with C4 architecture diagrams and code walkthrough for the current git changes

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

9 rules. 0 are common, 3 are distinctive, and 6 are very specific to this project.

  • 0

    Common

    Most of the 74 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 6

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Never write more than ~100 lines without running the project's test command.

    line 28
  2. Distinctive:

    Never mix unrelated concerns in one slice (feature + refactor + deps = separate slices).

    line 29
  3. Very specific:

    Never expand scope beyond the current task — note adjacent issues.

    line 30
  4. Very specific:

    Never skip verify because "you'll test at the end." After each slice: test → verify → commit (see git-workflow-and-versioning).

    line 31
  5. Very specific:

    After each slice: test → verify → commit (see git-workflow-and-versioning).

    line 32
  6. Distinctive:

    Read the approved plan or task.

    line 40
  7. Very specific:

    Run the project's test command.

    line 48
  8. Very specific:

    Commit with a descriptive message (git-workflow-and-versioning).

    line 52
  9. Very specific:

    After npm test passes → commit feat: add task creation endpoint and form.

    line 118

“Common” is measured: kept by at least 6 in 10 of the 74 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

Incremental Implementation

You ship multi-file work as a sequence of small, working slices. Each slice leaves the repo buildable and tests green.

Hard Rules

Never write more than ~100 lines without running the project's test command. Never mix unrelated concerns in one slice (feature + refactor + deps = separate slices). Never expand scope beyond the current task — note adjacent issues; do not fix them. Never skip verify because "you'll test at the end." After each slice: test → verify → commit (see git-workflow-and-versioning).


Workflow

Step 1 — Confirm slice plan

Read the approved plan or task. Pick one slice: the smallest end-to-end path (vertical slice preferred). State what's in scope and explicitly out of scope for this slice.

Step 2 — Implement one slice

Build only that slice. Prefer the simplest thing that could work — no abstractions before the third real use case.

Step 3 — Test and verify

Run the project's test command. Run build/typecheck/lint if the change could affect them. Do not re-run the same command on unchanged code.

Step 4 — Commit and pick next slice

Commit with a descriptive message (git-workflow-and-versioning). If the slice came from a saved plan, tick its checkbox in the plan file before moving on — the plan is the durable progress record across sessions. Repeat Steps 2–4 until the task is done.

Step 5 — End-to-end check

Identify the exact verify command (test/build/lint), run it fresh, and read its actual output — never report "done" or "tests pass" from memory or assumption. Full test suite, build clean, feature matches spec/plan. Report slices completed and commits made.


Slicing strategies

StrategyWhen
Vertical slice (default)One complete path through stack (e.g. create → API → UI)
Contract-firstParallel FE/BE — define API contract first, then implement each side
Risk-firstProve the riskiest integration before surrounding code

Gotchas

  • A bug in slice 1 invalidates slices 2–5 — test each slice.
  • "While I'm here" refactors belong in a separate commit, not this slice.
  • Feature flags beat long-lived branches for incomplete user-visible work.
  • Agents that skip verify to save time usually cost more time in debugging.

Common Rationalizations

ExcuseReality
"I'll test everything at the end"Failures compound; you won't know which slice broke.
"Faster to do it all at once"Until one of 500 lines is wrong and you can't bisect.
"Too small to commit"Small commits are free; giant commits are expensive to revert.
"I'll add the feature flag later"Incomplete features must not be user-visible without a flag.
"Let me run tests again for reassurance"Re-run only after code changed since last green run.
"It should pass now"Claims of done require a fresh command run and read output — not a prediction.

Output Format

## Incremental implementation — [task/slug]

Slices planned: N | Completed: N
Current slice: [name] — [in scope / out of scope]

Verification:
- Tests: [command] → [pass/fail]
- Build: [command] → [pass/fail/skip]

Commits: [list or "pending user commit policy"]

Next slice: [one line] | Done

Examples

<examples> <example> <input>Implement task creation from the plan — DB, API, and UI.</input> <output> Slice 1 only: POST /tasks + migration + minimal form. Out of scope: list/edit/delete. After `npm test` passes → commit `feat: add task creation endpoint and form`. Next slice: list tasks. </output> </example> </examples>

Verification

  • Slice scope stated before coding
  • Tests run for slice
  • Commit advised after verification
  • Traceability to plan tasks

Red Flags

  • Slice shipped without tests while later slices depend on it
  • While-I-am-here refactor bundled into feature slice
  • Long-lived branch used instead of flag for partial UX
  • Slice boundary unclear — spans unrelated subsystems

Prune Log

Last pruned: 2026-07-09

  • Added evidence-before-claims verify gate + plan-checkbox ticking (agent-loom Phase 4, obra/superpowers)

Impact Report

Task: [slug] | Slices: N/M | Last verify: [pass/fail]
Commits: [count] | Scope expansions noted: [count]

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.