dotclaude

dotclaude

Opinionated Claude Code plugin that shapes how Claude thinks about your product, writes your code, and reviews the result.

Install

# Add the repo as a marketplace source, then install
/plugin marketplace add realDragonium/dotclaude
/plugin install dotclaude@realDragonium-dotclaude

# Local development
claude --plugin-dir ./path-to-dotclaude

The Workflow

These skills form a pipeline. Use them in order, or pick what you need:

product-reasoning  →  feature-discovery  →  code-mode + test-writing  →  review
"should we build?"    "what, how, where?"    "build it right"             "did we build it right?"

Skills

/dotclaude:product-reasoning

Interrogates feature proposals before any planning starts. Forces assumption audits, jobs-to-be-done analysis, and honest comparison against "do nothing."

/dotclaude:product-reasoning Add real-time collaboration to the editor

/dotclaude:feature-discovery

Plans a feature through codebase discovery and iterative dialogue. Runs a feature-scout subagent to map the codebase first, then works through decisions one at a time (high-confidence suggestions vs. explicit trade-off questions), then compiles a decision spec. The spec feeds directly into test-writing and code-mode.

/dotclaude:feature-discovery User authentication with email + OAuth

/dotclaude:code-mode

Persistent mode — activate once and it shapes all code Claude writes for the session. Enforces naming (no junk suffixes), composition (no inheritance), functional patterns (pure core, side effects at boundaries), and simplicity (no premature abstraction).

/dotclaude:code-mode

/dotclaude:test-writing

Writes behavior-focused tests. Tests describe what the code does, not how. Pairs with feature-discovery spec output. Supports TypeScript/JavaScript, Python, and Go idioms.

/dotclaude:test-writing Write tests for src/auth/

/dotclaude:review

Code and architecture review in one pass. Finds real bugs, security issues, and philosophy violations at the code level; challenges simplicity, structure, naming, and design decisions at the architecture level. Language-specific checklists for Go, Python, and TypeScript. Never flags formatting.

/dotclaude:review Review the last commit
/dotclaude:review Review the architecture of src/

Philosophy

All skills share principles defined in skills/shared/philosophy.md:

• Simplicity as default — complexity must earn its place with a concrete, current problem
• Composition over inheritance — never in our own code; only for language/framework mandates
• Functional patterns — pure functions for core logic, side effects at boundaries only
• Data is transparent — public fields, valid by construction, no logic-free getters/setters
• Precise naming — no junk suffixes; name what the thing IS or DOES
• Core at the center — framework and persistence are outer layers; core logic has no framework imports
• Formatting is automated — never flagged in review; that's prettier/black/gofmt's job

Language-specific extensions live alongside the core philosophy:

File	Covers
philosophy-go.md	Flat package structure, small interfaces, explicit errors
philosophy-python.md	Layered architecture conventions, dataclasses, type hints
philosophy-typescript.md	Frontend patterns, types, async, no barrel files

Edit any philosophy file to evolve. All skills pick up changes automatically.

Structure

dotclaude/
├── skills/
│   ├── shared/
│   │   ├── philosophy.md              ← Core principles (edit this)
│   │   ├── philosophy-go.md           ← Go-specific extensions
│   │   ├── philosophy-python.md       ← Python-specific extensions
│   │   └── philosophy-typescript.md   ← TypeScript/JS extensions
│   ├── product-reasoning/
│   │   └── SKILL.md                   ← Challenge feature proposals
│   ├── feature-discovery/
│   │   └── SKILL.md                   ← Scout codebase, plan iteratively
│   ├── code-mode/
│   │   └── SKILL.md                   ← Persistent coding philosophy mode
│   ├── test-writing/
│   │   └── SKILL.md                   ← Behavior-focused test generation
│   └── review/
│       ├── SKILL.md                   ← Code + architecture review
│       └── references/
│           ├── typescript-checklist.md
│           ├── python-checklist.md
│           └── go-checklist.md
├── agents/
│   ├── feature-scout.md               ← Read-only codebase explorer
│   ├── product-critic.md
│   └── reviewer.md
└── README.md

Extending

Add a language checklist: create skills/review/references/<lang>-checklist.md, add detection in skills/review/SKILL.md.

Add a language philosophy: create skills/shared/philosophy-<lang>.md following the existing pattern. Add it to the detection list in philosophy.md.

Add a skill: create skills/<name>/SKILL.md with frontmatter. It auto-discovers.