convez.ai
Skill
camoa

claude-skills

camoa/claude-skills

Ver no GitHub
BomPluginScore · de 0 a 100

Instalação

$npx convez-ai camoa/claude-skills

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
33
Forks
6
Acompanhando
2
Saúde
28%
Último commit
há 1 semana
Versão
—
Issues abertas
5
Contribuidores
1

README

A tradução para o português deste README ainda está na fila.

Claude Skills

Custom plugins and tools for Claude Code.

Heads up (transition notice, kept through ~August 2026). drupal-dev-framework has been renamed to ai-dev-assistant (same workflow, broader scope, command names unchanged). If you already have the old plugin installed: install ai-dev-assistant, run /drupal-dev-framework:upgrade once to migrate your project store and per-project wiring, then uninstall drupal-dev-framework (/plugin uninstall drupal-dev-framework@camoa-skills). The shell exposes only /drupal-dev-framework:upgrade; everything else now lives under the ai-dev-assistant: namespace. A few other plugins got minor description and tone cleanups in the same pass. Full steps: drupal-dev-framework/README.md.

Background

I started building what I called "frameworks" over a year before Claude officially released Skills. Same concept, different name.

The idea came from frustration. I was tired of repeating the same instructions every conversation. Instead of starting fresh each time, I asked AI to analyze our successful interactions and create frameworks capturing recurring requirements and preferences. These became reusable project knowledge.

This approach produced results: 3 published Drupal contrib modules, 17+ blog articles, automated social media campaigns, and phase-based editorial workflows.

When Claude released Skills officially, I recognized what I'd been building. This repository translates those frameworks into proper Skills with tooling.

I wrote more about this methodology in My Journey with AI Tools.

Installation

# Add the marketplace
/plugin marketplace add https://github.com/camoa/claude-skills

# Install individual plugins
/plugin install dev-guides-navigator@camoa-skills
/plugin install ai-dev-assistant@camoa-skills
/plugin install plugin-creation-tools@camoa-skills
/plugin install brand-content-design@camoa-skills
/plugin install code-quality-tools@camoa-skills
/plugin install drupal-htmx@camoa-skills
/plugin install code-paper-test@camoa-skills
/plugin install drupal-ai-contrib@camoa-skills

Using these plugins outside Claude Code

Skills in this marketplace conform to the open agentskills.io standard and work in Cursor, Codex CLI, VS Code Copilot, Gemini CLI, Cline, OpenCode, and more. Commands, agents, and hooks are Claude-Code-specific by format but can be emulated. See PORTABILITY.md for the full guide and CURSOR.md for Cursor-specific notes.

Known Issues

Skills not auto-discovered on startup

There's a known bug in Claude Code affecting git-based marketplaces. Skills may fail to load during initialization with "no such file or directory" errors because Claude Code looks for skill files in the wrong location.

Workaround: Skills still work when invoked via the Skill tool (slash commands). The issue only affects automatic discovery at startup.

Status: Awaiting fix from Anthropic.

Plugins

dev-guides-navigator (v0.1.0)

Smart guide discovery and routing for the dev-guides site. Routes AI to the correct guide using hash-based caching and KG metadata for disambiguation.

  • Caches llms.txt with hash-based freshness check: no redundant fetches
  • KG metadata (concepts/not fields) prevents wrong-guide selection (e.g., "story.yml" routes to UI Patterns, not Storybook)
  • Two-hop routing: llms.txt → topic index.md → specific guide
  • 1200+ atomic decision guides across 66 topics
# Invoked automatically before design/dev tasks, or manually:
/dev-guides-navigator style guide

ai-dev-assistant (v5.0.0)

An AI assistant for developers that focuses on getting the process right, not just getting code out fast. Most AI dev tools optimize for speed. This one keeps the work disciplined: understand the problem before coding, reuse what already exists, follow your standards, and verify. It runs each task through Research → Architecture → Implementation → Review, with deterministic gates (SOLID, TDD, DRY, security, code purposefulness) the AI can't quietly skip, and grounds its decisions in best-practice guides instead of whatever the model guessed. Requires dev-guides-navigator. Stack-agnostic engine; ships with a Drupal-flavored reference implementation for the deep components (a stack-neutral generalization is in progress).

Renamed from drupal-dev-framework. Already using the old plugin? Install ai-dev-assistant, then run /drupal-dev-framework:upgrade once from the deprecated shell: it migrates your project store and per-project hooks to the new name, after which the old plugin is safe to uninstall. The shell deliberately exposes only /drupal-dev-framework:upgrade (all old command names route to it); run everything else under the unchanged-name ai-dev-assistant: namespace. See drupal-dev-framework/README.md for the disable/uninstall steps.

/plugin install dev-guides-navigator@camoa-skills   # Required dependency
/plugin install ai-dev-assistant@camoa-skills

/ai-dev-assistant:new my_project        # Create project
/ai-dev-assistant:next                  # Continue work (main entry point)
ComponentContents
Commands44: the full lifecycle (/new, /next, /research, /research-team, /design, /implement, /complete, /review, /validate*, work-orders, visual/E2E review, playbooks, worktrees)
Agents10 with cost control: project-orchestrator, architecture-drafter, architecture-validator (isolated worktree), pattern-recommender, contrib-researcher, journey-discovery, guides-matcher, analysis, AI test selector, work-order critic
Skills23 (phase management, TDD companion, guide integration, work-order compiler/loop, context loading)
Referencesmethodology docs (SOLID, TDD, DRY, Library-First, Quality Gates, Purposeful Code) + contracts/walkthroughs
HooksSessionStart (dependency check + project context), PreCompact (context preservation)

Features competing agent research (/research-team) with Build/Use/Extend debate for features and competing hypothesis investigation for bugs.

See ai-dev-assistant/README.md for full documentation.

plugin-creation-tools (v2.3.0)

Complete guide for creating Claude Code plugins: skills, commands, agents, hooks, MCP servers, and configuration. Covers 18 hook events, 4 hook types (command/prompt/agent/HTTP), agent isolation and cost control, marketplace distribution with 6 source types, and pushy description optimization.

# Just describe what you want:
Create a plugin called "my-tools" with a deploy command

# Or use specific commands:
/plugin-creation-tools:create my-tools --skill --agent --hook
/plugin-creation-tools:validate ./my-tools
ComponentContents
Skills1 (plugin-creation: 30+ reference docs, templates, examples)
Commands3 (/create, /add-component, /validate)
Agents2 (skill-quality-reviewer, plugin-structure-auditor)

brand-content-design (v2.8.0)

Create branded presentations, LinkedIn carousels, infographics, and HTML pages with consistent visual identity.

/brand-content-design:brand-init        # Start a new brand project
ComponentContents
Commands19 (brand management, templates, content creation: quick and guided modes)
Skills4 (brand-content-design, visual-content, infographic-generator, html-generator)
Agents1 (brand-analyst with project memory)

Features 21 visual styles, 114 infographic templates, HTML design systems with 15 component types, and 17 color palette types.

See brand-content-design/README.md for full documentation.

code-quality-tools (v2.7.0)

Code quality and security auditing for Drupal (via DDEV) and Next.js projects.

/code-quality-tools:audit               # Full audit with synthesis
/code-quality-tools:review              # Rubric-scored code review
/code-quality-tools:security-debate     # 3-agent security debate
/code-quality-tools:architecture-debate # 3-agent architecture debate
ComponentContents
Commands11 (/setup, /audit, /review, /coverage, /security, /security-debate, /architecture-debate, /lint, /solid, /dry, /tdd)
Skills1 (code-quality-audit)

Security coverage: Drupal (10 layers: PHPStan, Psalm, PHPMD, Semgrep, Trivy, Gitleaks, Roave, Drush, Composer audit, Security Review) and Next.js (7 layers). Optional DAST with OWASP ZAP + Nuclei.

Features rubric-scored code review (/50 with quality gate), cross-audit correlation with prioritized action plans, and two 3-agent debate commands with isolated worktrees.

See code-quality-tools/README.md for full documentation.

drupal-htmx (v1.4.0)

HTMX development guidance and AJAX-to-HTMX migration tools for Drupal 11.3+.

ComponentContents
Commands5 (/htmx, /htmx-analyze, /htmx-migrate, /htmx-pattern, /htmx-validate)
Agents3 (ajax-analyzer, htmx-recommender, htmx-validator)
Skills1 (htmx-development)
References4 (quick-reference, htmx-implementation, migration-patterns, ajax-reference)

See drupal-htmx/README.md for full documentation.

code-paper-test (v0.4.0)

Systematically test code, skills, commands, and configs through mental execution: trace logic line-by-line with concrete values to find bugs, AI hallucinations, edge cases, and contract violations before deployment.

ComponentContents
Skills1 (paper-test: with data flow tracking, error propagation, config validation, severity scoring)
Commands1 (/test-team: competing agent team with isolated worktrees: Happy Path + Edge Case + Red Team)
References11 (core method, dependencies, contracts, AI auditing, hybrid testing, common flaws, advanced techniques + state machines, severity scoring, blind A/B comparison, rubric scoring, skill/config testing)

Tests code AND non-code artifacts: skills, commands, agents, YAML configs. Auto-detects skill files and switches to instruction tracing mode.

See code-paper-test/README.md for full documentation.

drupal-ai-contrib (v0.1.0)

AI-assisted Drupal contribution quality (evidence over assertion): every gate passes only on a produced, captured artifact, never on an AI assertion. Mirrors the drupalci pipeline locally at CI strictness, gates on the adopted AI-contribution policy, reviews work in fresh-context agents, and confirms the real GitLab pipeline.

/drupal-ai-contrib:setup     # Onboard + environment-match a contribution workspace
/drupal-ai-contrib:verify    # Local drupalci-parity + AI-policy + eval gates
ComponentContents
Commands6 (/setup, /issue, /verify, /review, /submit, /pipeline): the detect-driven contribution arc
Skills8 (drupal-ai-contrib umbrella/router, 6 worker skills, contribution-guardrails discipline)
Agents3 read-only (fresh-context-reviewer, external-fact-verifier, ai-policy-checker)
HooksPostToolUse (re-verification ledger), SessionStart (contribution-workspace reminder)

The drupalci-parity gate set (composer/phpcs/phpstan/phpunit/cspell/eslint/stylelint) mirrors each enabled gitlab_templates job at its real strictness. Cites the camoa/dev-guides contribution guides by slug via dev-guides-navigator; a contribution is run as an ai-dev-assistant task.

See drupal-ai-contrib/README.md for full documentation.

Plugin Conventions

All plugins follow these conventions:

  • CLAUDE.md at plugin root: plugin-specific rules and conventions
  • .claude/rules/: path-scoped convention files
  • Frontmatter standards: version, model routing (haiku/sonnet/opus), disallowedTools on read-only agents
  • Progressive disclosure: SKILL.md under 500 lines; details in references/
  • Imperative voice: skills and commands give instructions, not documentation

Official Documentation

Related Tools

ToolPurposeInstall
Anthropic skill-creatorOfficial skill creator/plugin install example-skills@anthropic-agent-skills
SuperpowersTDD skill development/plugin marketplace add obra/superpowers-marketplace
Skill Seeker MCPAutomated doc scrapingSee repo README

Acknowledgments

This marketplace was built collaboratively by Carlos Ospina and Claude (Anthropic). The methodology, frameworks, and domain expertise came from Carlos. Claude contributed code generation, research synthesis, documentation structure, and pattern implementation.

Patterns and insights drawn from:

License

MIT

Ver o original no GitHub

Relacionados