sethdford/claude-standards

Dernière activité source enregistrée: 11 mars 2026 à 12:03

Catalogue SkillsMP mis à jour: 12 novembre 2026 à 12:52

skills collectés: 8

Étoiles GitHub: 0
Forks GitHub: 0

Installer avec un assistant IA

Copiez ce prompt dans l’assistant IA que vous utilisez.

Follow https://skillsmp.com/skill-install/prompt.md to install Agent Skills from https://github.com/sethdford/claude-standards.

README du dépôt

Fichier source du README : README.md

Lire le README source

Afficher le README complet

# claude-standards Standards-driven agent development framework. Define, distribute, and enforce project standards so AI agents and humans share the same rules and don't drift from them over time. ## The Problem: Agentic Standards Drift When agents work without clear standards, they gradually diverge from project conventions. Code style drifts. Architecture decisions get made inconsistently. AI insights become unreliable. The longer this continues, the harder it is to fix. ## The Solution: Graduated Standards Enforcement claude-standards provides a maturity model with three levels. Start simple, add rigor as your project grows. ``` L1: Context (Passive) → Standards loaded into agent context L2: Ceremonies (Scheduled) → Drift caught at checkpoints L3: Enforcement (Active) → Drift caught in real-time ``` ## Quick Start ### L1: Get Standards into Agent Context (hours) 1. Copy `levels/L1-context/scaffold/docs/standards/` into your project 2. Copy reference standards from `levels/L1-context/library/` that apply to your stack 3. Customize `[CUSTOMIZE]` markers in each standard 4. Copy and adapt `levels/L1-context/templates/CLAUDE.md.template` → `CLAUDE.md` 5. Copy and adapt `levels/L1-context/templates/AGENTS.md.template` → `AGENTS.md` > **Important:** After copying any template, you MUST replace all `[PROJECT_NAME]`, `[CUSTOMIZE]`, and `[placeholder]` sections with your actual project values. Templates used as-is without customization will cause agents to load placeholder text as "standards." See `levels/L1-context/README.md` for detailed instructions. ### L2: Add Drift Detection Ceremonies (days) 1. Copy scripts from `levels/L2-ceremonies/scripts/` → `scripts/` in your project 2. Copy ceremony definitions from `levels/L2-ceremonies/ceremonies/` 3. Copy GitHub templates from `levels/L2-ceremonies/github/` 4. Run the weekly drift audit every Monday See `levels/L2-ceremonies/README.md` for detailed instructions. ### L3: Add Real-Time Enforcement (week+) 1. Copy agents from `levels/L3-enforcement/agents/` → `.claude/agents/` 2. Copy commands from `levels/L3-enforcement/commands/` → `.claude/commands/` 3. Install hooks from `levels/L3-enforcement/hooks/` 4. Customize the file-to-standards mapping in standards-enforcer.md See `levels/L3-enforcement/README.md` for detailed instructions. ## Install via Claude ```bash # Install L1 skills (write, index, wire standards) claude install github:sethdford/claude-standards/plugins/standards-authoring # Install L2 skills (drift audits, PR gates, ceremonies) claude install github:sethdford/claude-standards/plugins/standards-ceremonies # Install L3 skills (enforcement agents, hook design) claude install github:sethdford/claude-standards/plugins/standards-enforcement ``` ## Reference Library The `levels/L1-context/library/` contains generic, adoptable standards extracted from real production projects: | Category | Standards | | ------------- | --------------------------------------------------------------------------------------------------------------------------- | | ai/ | hallucination-prevention, prompt-engineering, insight-quality, citation-and-sourcing, agent-architecture, human-in-the-loop | | engineering/ | typescript, react-nextjs, api-design, database, testing, workflow, performance | | quality/ | governance, ceremonies, code-review, security, performance, benchmarking | | design/ | design-system, accessibility, responsive, animation, data-visualization | | operations/ | deployment, monitoring, incident-response | | architecture/ | monorepo-structure, package-boundaries | | analytics/ | methodology, metrics-dictionary, event-tracking, benchmarking-methodology | ## Methodology See `methodology/` for the core philosophy: - `manifesto.md` — Five core principles - `maturity-model.md` — Detailed level descriptions and graduation criteria - `anti-patterns.md` — Seven common failures and how to avoid them ## Reference Implementation The [aim monorepo](examples/aim/README.md) shows what this looks like in a real production project using all three levels. ## Contributing See [CONTRIBUTING.md](CONTRIBUTING.md). ## License MIT — Seth Ford

Chargement du README du dépôt…