Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/nathanstaines-codebase-design
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: design
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/codebase-design
What's inside
- Skill name
codebase-design- Description
- Shared vocabulary for designing deep modules. Use when the user wants to design a module's interface or place its seam, or to find deepening opportunities. Also use when another skill needs the deep-module vocabulary.
- License
- MIT
- Instructions
- 122 lines in SKILL.md · view on GitHub
Files (4)
- DEEPENING.md
- DESIGN-IT-TWICE.md
- SKILL.md
- openai.yaml · agents/
SKILL.md (first lines)
# Codebase design
Design **deep modules**: a lot of behaviour behind a small interface, placed at a clean seam, verifiable through that interface. Use this language and these principles wherever something with an interface is being designed or restructured, whatever it's made of: imperative code, stylesheets, design tokens, configuration. The aim is leverage for callers, locality for maintainers and verifiability for everyone.
"Caller" throughout means any consumer of the interface: code that calls a function, a stylesheet that `@use`s a partial, markup that applies a class, a build that reads a config.
## Glossary
Use these terms exactly; don't substitute "component", "service", "API" or "boundary". Consistent language is the whole point.
**Module**: anything with an interface and an implementation. Deliberately scale-agnostic and substance-agnostic: a function, class, package, SCSS partial, design-token layer or tier-spanning slice. _Avoid_: unit, component, service.
**Interface**: everything a caller must know to use the module correctly: the type signature, but also invariants, ordering constraints, error modes, required configuration and performance characteristics. In declarative code that includes the mixins, functions, variables and custom properties exposed, plus whatever markup structure, cascade position or load order the module assumes. _Avoid_: API, signature (too narrow; they name only the type-level surface).
**Implementation**: what's inside a module, its body of code. Distinct from **adapter**: a thing can be a small adapter with a large implementation (a Postgres repository) or a large adapter with a small implementation (an in-memory fake). Reach for "adapter" when the seam is the topic; "implementation" otherwise.
**Depth**: leverage at the interface: the amount of behaviour a caller (or check) can exercise per unit of interface they have to learn. A module is **deep** when a large amount of behaviour sits behind a small interface, **shallow** when the interface is nearly as complex as the implementation.
**Seam** _(Michael Feathers)_: a place where you can alter behaviour without editing in that place; the *location* at which a module's interface lives. Where to put the seam is its own design decision, distinct from what goes behind it. _Avoid_: boundary (overloaded with DDD's bounded context).
**Adapter**: a concrete thing that satisfies an interface at a seam. Describes *role* (what slot it fills), not substance (what's inside). A theme satisfying a token contract is as much an adapter as an HTTP client satisfying a port.
**Leverage**: what callers get from depth: more capability per unit of interface they learn. One implementation pays back across N call sites and M checks.
**Locality**: what maintainers get from depth: change, bugs, knowledge and verification concentrate in one place rather than spreading across callers. Fix once, fixed everywhere.
## Deep vs shallow
**Deep**:
```
┌─────────────────────┐
│ Small interface │ ← Few entry points, simple params
├─────────────────────┤
│ │
│ Deep implementation │ ← Complex logic hidden
│ │
└─────────────────────┘
```
**Shallow**:
```
┌─────────────────────────────────┐
│ Large interface │ ← Many entry points, complex params
├─────────────────────────────────┤
│ Thin implementation │ ← Just passes through
└─────────────────────────────────┘
```
Depth is substance-agnostic. A `pagination($total, $current)` mixin that hides layout, states and responsive behaviour is deep; a `$button-border-radius` variable that renames one CSS property is shallow. A `parseSchedule(text)` function that hides a grammar is deep; a `getName()` wrapper over `.name` is shallow.
When designing an interface, ask:
- Can I reduce the number of entry points?
- Can I simplify the parameters?
- Can I hide more complexity inside?
## Principles
- **Depth is a property of the interface, not the implementation.** A deep module can be internally composed of small, swappable parts; they just aren't part of the interface. A module can have **internal seams** (private to its implementation, used by its own checks) as well as the **external seam** at its interface.
- **The deletion test.** Imagine deleting the module. If complexity vanishes, it was a pass-through. If complexity reappears across N callers, it was earning its keep.
- **The interface is the verification surface.** Callers and checks cross the same seam, whether the check is a unit test calling a function, a snapshot of compiled output, visual regression over rendered states or a manual verification flow. If verifying means reaching past the interface, the module is probably the wrong shape.
- **One adapter means a hypothetical seam. Two adapters means a real one.** Don't introduce a seam unless something actually varies across it.
- **Prefer existing seams, and the highest one available.** Reach for a seam that already exists before cutting a new one, and when a new one is needed, propose it at the highest point you can. The fewer seams across the codebase, the better; the ideal number is one.
## Designing for verifiability
Good interfaces make verification natural:
1. **Accept dependencies, don't create them.**
```typescript
// Verifiable
function processOrder(order, paymentGateway) {}
// Hard to verify
function processOrder(order) {
const gateway = new StripeGateway();
}
```
```scss
// Verifiable: the theme enters at the seam
@mixin badge($palette) { /* ... */ }
// Hard to vary: reaches past the seam for a global
@mixin badge() { color: $brand-colour; }
```
2. **Return results, don't produce side effects.**
```typescript
// Verifiable
function calculateDiscount(cart): Discount {}
// Hard to verify
function applyDiscount(cart): void {
cart.total -= discount;
}
```
```scss
// Verifiable: emits CSS only where included
@mixin visually-hidden { /* ... */ }
// Side effect: emits CSS the moment the file is @used
* { box-sizing: border-box; }
```
3. **Small surface area.** Every entry point is another thing to check; every parameter is more setup per check.
## Rejected framings
- **Depth as a ratio of implementation lines to interface lines** (Ousterhout): rewards padding the implementation. Use depth-as-leverage instead.
## Going deeper
- **Deepening a cluster given its dependencies**: see [DEEPENING.md](./DEEPENING.md) for dependency categories, seam discipline and replace-don't-layer verification.From nathanstaines/skills, MIT.
More from nathanstaines/skills
- 0.565score
tdd
nathanstaines
Test-driven development, the red → green loop.
SkillAFreshness: AClaude Code +9 - 0.565score
code-review
nathanstaines
Two-axis review of the changes since a fixed point, Standards and Spec.
SkillAFreshness: AClaude Code +9 - 0.561score
architecture-review
nathanstaines
Scan a codebase, or a named area of it, for deepening opportunities, present them as a visual HTML report, then triage every candidate yes/maybe-later/no and…
SkillAFreshness: AClaude Code +9 - 0.561score
progress
nathanstaines
Show the state of every feature in the task tracker at a glance: what's done, what's in flight, what's startable now and which specs are still waiting on a…
SkillAFreshness: AClaude Code +9 - 0.561score
to-tasks
nathanstaines
Break a plan, spec or the current conversation into a set of tracer-bullet tasks, each declaring its blocking edges, published to the configured task tracker.
SkillAFreshness: AClaude Code +9 - 0.561score
scout
nathanstaines
Chart a large, foggy piece of work as a map of decision tasks on the project's task tracker, then resolve them one at a time until the way to a spec is clear.
SkillAFreshness: AClaude Code +9 - 0.561score
setup-skills
nathanstaines
Configure this project for engineering skills.
SkillAFreshness: AClaude Code +9 - 0.558score
retro
nathanstaines
Review a coding session and propose improvements to the agent's environment for future runs.
SkillAFreshness: AClaude Code +9 - 0.553score
implement
nathanstaines
Implement a piece of work described by a spec or set of tasks.
SkillAFreshness: AClaude Code +9 - 0.544score
diagnose-bugs
nathanstaines
Diagnosis loop for hard bugs and performance regressions.
SkillAFreshness: AClaude Code +9 - 0.539score
domain-modelling
nathanstaines
Build and sharpen a project's domain model.
SkillAFreshness: AClaude Code +9 - 0.539score
prototype
nathanstaines
Build a throwaway prototype to answer a design question.
SkillAFreshness: AClaude Code +9
Signals
- Stars
- 0
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Sep 30, 2026 (1 month ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/nathanstaines/skills
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from nathanstaines
- 0.565score
tdd
nathanstaines
Test-driven development, the red → green loop.
SkillAFreshness: AClaude Code +9 - 0.565score
code-review
nathanstaines
Two-axis review of the changes since a fixed point, Standards and Spec.
SkillAFreshness: AClaude Code +9 - 0.561score
architecture-review
nathanstaines
Scan a codebase, or a named area of it, for deepening opportunities, present them as a visual HTML report, then triage every candidate yes/maybe-later/no and…
SkillAFreshness: AClaude Code +9 - 0.561score
progress
nathanstaines
Show the state of every feature in the task tracker at a glance: what's done, what's in flight, what's startable now and which specs are still waiting on a…
SkillAFreshness: AClaude Code +9 - 0.561score
to-tasks
nathanstaines
Break a plan, spec or the current conversation into a set of tracer-bullet tasks, each declaring its blocking edges, published to the configured task tracker.
SkillAFreshness: AClaude Code +9 - 0.561score
scout
nathanstaines
Chart a large, foggy piece of work as a map of decision tasks on the project's task tracker, then resolve them one at a time until the way to a spec is clear.
SkillAFreshness: AClaude Code +9
Cite this page
codebase-design skill. Agent Mart. https://agentmart.dev/skills/nathanstaines-codebase-design (last updated 2026-10-24)