Agent skill

core-workflow

Quellcode ansehen: travisjneuman/.claude

#116Globales Ranking · von 116 SkillsSafe

Installation

npx skills add travisjneuman/.claude --skill core-workflow

137

Installationen

EU-hosted inference API

Power your AI agent skills with open-source models.

Drop-in OpenAI-compatible API. No data leaves Europe.

MiniMax

MiniMax M3

$0.50 / $1.90

per M tokens

Z.ai

GLM 5.3 Flash

$0.20 / $0.60

per M tokens

MoonshotAI

Kimi K3

$4.00 / $18.00

per M tokens

DeepSeek

DeepSeek V4.1 Flash

$0.40 / $1.42

per M tokens

Core Workflow Patterns

Comprehensive development workflow reference. This loads on-demand for detailed patterns.

Quick links to rules: ~/.claude/rules/ for stack-specific and task-type checklists.


Session Protocol

Start Checklist

# 1. Sync first: default branch from origin/HEAD, fetch + prune, fast-forward
git fetch --prune origin && git merge --ff-only "$(git symbolic-ref --short refs/remotes/origin/HEAD)"

# 2. Get context
git log -3

# 3. Check for existing work
ls tasks/*.md 2>/dev/null || echo "No active tasks"

Take today's date from the environment context rather than inferring it.

End Checklist

# 1. Verify by review: re-read the request and the whole diff (no test/type-check ladder)

# 2. Archive completed work
mv tasks/<completed-task>.md .archive/completed-tasks/

# 3. Commit with comprehensive message
git add -A  # all pending changes, any session's
git commit  # See Git Conventions below
git push    # or the configured push runner, when one is defined; never force

Stop any dev server you started: lsof -ti:PORT | xargs kill (or Windows equivalent)


Sizing Work

Complexity Workflow
Simple fix (<30 min) Direct execution
Single feature (30min-2hr) Task file + TodoWrite
Multi-phase feature (days) Plan mode, then one task file per phase in tasks/
New project/app /init-project, then plan mode for the first phase

Git Conventions

Commit Types

feat | fix | refactor | perf | test | docs | chore

Commit Message Format

type: Short summary (50 chars max)

## What Changed
- File X: Added feature Y
- File Z: Updated config for A

## Why
- User requested feature Y
- Config A needed update

## Verification
- Diff reviewed against the request, callers, and edge cases

Auto-Commit on Task Completion

When a task or plan is complete, automatically commit without being asked.

Pre-Commit Checks

# 1. Check this is a user-owned repo (not external)
git remote get-url origin | grep -q "travisjneuman" && echo "OK: User repo"

# 2. Check push is not blocked
git remote get-url --push origin | grep -q "no_push" && echo "SKIP: External repo"

Rules

Condition Action
User's own repo Auto-commit + push
External repo (no_push) Never commit - read-only
Submodule (external) Never commit - read-only
Uncommitted secrets detected Block - warn user

Security Checklist

Frontend

  • textContent not innerHTML
  • unknown type for external data
  • No exposed API keys
  • HTTPS for external requests
  • Input sanitization

Backend

  • Input validation on all endpoints
  • Auth guards on protected routes
  • Parameterized queries (no raw SQL)
  • Secrets in environment variables

Performance Targets

Metric Target
Initial bundle <100KB gzipped
Page load <1s
Interaction latency <100ms
Lighthouse Performance 95+
Accessibility WCAG AA minimum

Accessibility (WCAG AA)

  • Semantic HTML structure
  • Alt text for meaningful images
  • Keyboard navigation (Tab, Enter, Escape)
  • Focus indicators visible
  • Color contrast >= 4.5:1
  • ARIA labels on interactive elements
  • Touch targets >= 44x44px

Thinking Frameworks

Use structured decision-making for complex choices:

Decision Type Framework
Long-term implications /consider:10-10-10
Root cause analysis /consider:5-whys
Prioritization /consider:eisenhower-matrix
Innovation /consider:first-principles
Risk identification /consider:inversion
Simplification /consider:occams-razor
Focus /consider:one-thing
Tradeoffs /consider:opportunity-cost
Optimization /consider:pareto
Consequences /consider:second-order

Debugging Protocol

Standard Issues

  1. Reproduce the issue
  2. Read relevant code
  3. Identify root cause
  4. Fix the root cause
  5. Verify by review of the diff and the output the fix already produces

Intermittent/Complex Issues

Use the debug-systematic skill for a systematic approach.


Build vs Buy Philosophy

We build features. We use utilities.

  • Build: All feature logic, business rules, UI/UX, data models
  • Use: Low-level abstractions (D3, Recharts, Lexical, Konva)
  • Criterion: We own the feature, library handles complexity

License Requirements

  • Must use: MIT, Apache 2.0, BSD
  • Never use: GPL, AGPL (blocks commercialization)

Communication Standards

Progress Updates

Give high-level updates, not spam:

"Added authentication middleware (3 files) and updated the user store with the new
fields; now reviewing the login flow."

When to Ask

Use AskUserQuestion only when:

  • An essential choice is missing and changes the outcome
  • Instructions conflict and precedence doesn't settle it
  • A destructive or external target can't be identified exactly
  • The step needs authority the request didn't give

Otherwise pick the reasonable option, proceed, and state the assumption.

Directness Protocol

  • Logic over feelings
  • Correctness over validation
  • Direct feedback over diplomacy
  • Best solution over agreement

Context Hygiene

Reduce Token Usage

  • Short, high-signal summaries over long logs
  • Don't @-embed large docs by default
  • Reference paths + when to read them
  • Use /clear after completing work units

Delegation Patterns

Situation Action
Context >100k tokens Create prompt → delegate to fresh context
Moderate complexity /create-prompt → /run-prompt
Multi-stage features /create-meta-prompt
Approaching limits /whats-next for handoff document

Quick Reference

Common Commands

npm run dev          # Start dev server
npm run build        # Production build
npm run test         # Run tests (only when the user asks)
npm run type-check   # TypeScript check (only when the user asks)

File Naming

Type Convention Example
Components PascalCase UserCard.tsx
Hooks use prefix useAuth.ts
Utilities camelCase utils.ts
Tests .test.ts utils.test.ts

Resources

Official

Community


See Also

  • ~/.claude/docs/reference/checklists/ - Task-type specific checklists
  • ~/.claude/docs/reference/stacks/ - Stack-specific patterns
  • ~/.claude/docs/reference/tooling/ - Tool configuration guides
  • ~/.claude/skills/MASTER_INDEX.md - Full skills catalog
  • ~/.claude/agents/README.md - Agents directory

So verwenden Sie diesen Skill

  1. 1

    Install core-workflow by running npx skills add travisjneuman/.claude --skill core-workflow in your project directory. Führen Sie den obigen Installationsbefehl in Ihrem Projektverzeichnis aus. Die Skill-Datei wird von GitHub heruntergeladen und in Ihrem Projekt platziert.

  2. 2

    Keine Konfiguration erforderlich. Ihr KI-Agent (Claude Code, Cursor, Windsurf usw.) erkennt installierte Skills automatisch und nutzt sie als Kontext bei der Code-Generierung.

  3. 3

    Der Skill verbessert das Verständnis Ihres Agenten für core-workflow, und hilft ihm, etablierte Muster zu befolgen, häufige Fehler zu vermeiden und produktionsreifen Code zu erzeugen.

Was Sie erhalten

Skills sind Klartext-Anweisungsdateien — kein ausführbarer Code. Sie kodieren Expertenwissen über Frameworks, Sprachen oder Tools, das Ihr KI-Agent liest, um seine Ausgabe zu verbessern. Das bedeutet null Laufzeit-Overhead, keine Abhängigkeitskonflikte und volle Transparenz: Sie können jede Anweisung vor der Installation lesen und prüfen.

Kompatibilität

Dieser Skill funktioniert mit jedem KI-Coding-Agenten, der das skills.sh-Format unterstützt, einschließlich Claude Code (Anthropic), Cursor, Windsurf, Cline, Aider und anderen Tools, die projektbezogene Kontextdateien lesen. Skills sind auf Transportebene framework-agnostisch — der Inhalt bestimmt, für welche Sprache oder welches Framework er gilt.

Data sourced from the skills.sh registry and GitHub. Install counts and security audits are updated regularly.