simplicity-ladder
Skill1 starsChanged 28 days agoMITView on GitHub
Subagent-side, preloaded into the developer. How to pick the smallest correct implementation at Green — trace the flow first, climb the ladder (needed at all, already in repo, stdlib, platform, installed dependency, one line, minimum code), fix bugs at the shared root, and mark deliberate shortcuts.
In this repository20 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- adversarial-reviewSkill
- clean-architectureSkill
- decision-recordingSkill
- design-system-checkSkill
- dispatch-handoffSkill
- feature-branchingSkill
- finding-dispositionSkill
- git-recoverySkill
- gotcha-recordingSkill
- human-checkpointSkill
- plan-reviewSkill
- plan-writingSkill
- pr-createSkill
- spec-writingSkill
- subagent-contractSkill
- subagent-dispatchSkill
- tdd-loopSkill
- update-toolkitSkill
- wiki-updateSkill
The file
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 6 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈6
Distinctive · estimate
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
Every file the case touches and every caller of modified functions (trace the flow end to end).
- Distinctive: line 21
Standard library?: Use it.
- Distinctive: line 23
Already-installed dependency?: Use it.
- Distinctive: line 24
One line?: Write one line.
- Distinctive: line 25
Only then: write the minimum code that passes the test.
- Distinctive: line 26
New dependency = last resort: Never add one for what a few lines do.
≈ “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.
Files
Simplicity Ladder
Runs at Green (tdd-loop § Green, step 1) before writing production code, and before adding any file, helper, abstraction, or dependency.
Read first
- Every file the case touches and every caller of modified functions (trace the flow end to end).
architecture.md § Layers,§ Conventions, and the dependency manifest.
Steps
- Climb and stop at the first rung that holds:
- Does the test force it? Skip anything the case does not ask for and list it under
skipped:. - Already in this codebase? Grep for existing helpers, types, or patterns and reuse them.
- Standard library? Use it.
- Platform feature? (
<input type="date">, CSS, DB constraint, etc.) Use it. - Already-installed dependency? Use it.
- One line? Write one line.
- Only then write the minimum code that passes the test.
- Does the test force it? Skip anything the case does not ask for and list it under
- New dependency = last resort: Never add one for what a few lines do. If added, state in the report why rungs 2–5 fell short.
- Bug fix = root cause: Grep all callers before editing; fix the shared function rather than patching one caller.
- Mark deliberate shortcuts that have a known ceiling (
O(n²)scan, coarse lock, naive heuristic) with a code commentshortcut: <ceiling>; revisit when <trigger>and report it. Shortcuts without a trigger are forbidden. - Report skipped items as one line each:
skipped: <X> — add when <Y>.
Never simplify away
Trust-boundary input validation, data-loss error handling, security, accessibility, Behavior cases, required § Layers ports (rule 23), or the failing test first (rule 2).
Anti-patterns
- Single-implementation interfaces (ports excepted), single-product factories, unrequested config/scaffolding.
- Duplicating an existing helper.
- Defending simplifications with long prose (one
skipped:line suffices).
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.

