convez.ai
Skill
Claudfather

clauDNA

Claudfather/clauDNA

Ver no GitHub
InicialPluginScore · de 0 a 100

Global Claude Code configuration — skills, agents, hooks, commands installed to ~/.claude/

Instalação

$npx convez-ai Claudfather/clauDNA

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
0
Forks
1
Acompanhando
0
Saúde
75%
Último commit
há 2 dias
Versão
v0.35.4
Issues abertas
145
Contribuidores
1

README

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

clauDNA

Claude Code plugin pack: curated skills, agents, and hooks distributed via the Claudfather marketplace.

Quick Start

Requires Claude Code (npm install -g @anthropic-ai/claude-code).

Inside Claude Code:

/plugin marketplace add Claudfather/clauDNA
/plugin install claudna@Claudfather

That's it. Skills become available as /claudna:<skill-name> (namespaced under the plugin). Tab completion works — type /claudna: and press Tab to browse, or /claudna:tech and Tab to narrow. Hooks (auto-format, pretooluse permission expansion, notifications) activate automatically when the plugin is enabled.

After install

Optional, in any order:

  1. Recommended settings.json tweaks — permissions, statusLine, sandbox config. See SETUP_GUIDE §3.
  2. Snowflake setup — only if you use the Snowflake skills. See SETUP_GUIDE §5.
  3. Shell aliases — power-user worktree commands. See SETUP_GUIDE §6.
  4. Project setup — in any project, run /claudna:init-project to generate CLAUDE.md, CHANGELOG.md, and .claude/lessons.md.

Headless / CI / Docker

For bots, CI runners, and Docker images, drop a settings.json with enabledPlugins + extraKnownMarketplaces, then set CLAUDE_CODE_SYNC_PLUGIN_INSTALL=1 and run claude -p. Full recipe in SETUP_GUIDE §4.

What ships with the plugin

DirectoryCountContents
skills/52User-invocable slash commands
agents/8snowflake-analyst, dbt-engineer, neon-analyst, modal-ops, railway-ops, vercel-ops, code-reviewer, spec-reviewer
plugin-hooks/3 active + 1 opt-inAuto-format on Write/Edit, pretooluse permission expansion, macOS notifications. (statusline.sh is opt-in — see SETUP_GUIDE §3.2.) Named plugin-hooks/ to avoid a Claude Code bug that deletes any project-root hooks/ directory between tool calls.

Skills

Invocable as /claudna:<name> after marketplace install (e.g. /claudna:tech-debt). Bare names below are shown without the prefix for readability.

Workflow

SkillDescription
/session-resumePick up where you left off — scans git, PRs, plans, handoff notes
/session-handoffEnd-of-session capture — decisions, learnings, next steps
/repo-healthMulti-repo dashboard — open PRs, CI status, stale branches
/implement-planExecute a design doc or GitHub Issue — browse, pick, challenge, build, test, PR
/worktreeManage git worktrees for parallel Claude sessions
/development-retroPost-merge retrospective — systemic patterns, friction, breadcrumb trails
/name-sessionLabel the current session so it shows up cleanly in /resume
/weigh-development-pathsCompare multiple viable approaches at a development junction
/adversarial-reviewStructured plan challenge with evidence and anti-groupthink guards

Code Quality

SkillDescription
/tech-debtScan for tech debt, generate phased remediation plans
/security-auditScan 8 vulnerability categories, generate remediation plans
/review-prStructured PR review with severity-categorized findings
/review-changesReview uncommitted changes for bugs, conventions, security
/lessonsCapture and review lessons from corrections
/verify-completionVerify work is actually complete before claiming success
/frontend-performance-auditTrace re-render cascades, identify fetch/state/observer issues
/access-path-auditAudit cross-cutting concern consistency across system interfaces (API, CLI, MCP, Slack)
/visual-crawlAutonomous visual crawl — discover routes, screenshot at 3 viewports, file issues for findings

Product

SkillDescription
/product-enhanceDiscover enhancements via intent interviews and gap analysis
/product-visionArchitecture-aware product vision — explore what the codebase could become
/design-reviewVisual audit at 3 breakpoints with phased fix plans
/docs-reviewAudit docs for accuracy, update plan statuses, archive stale docs
/data-model-auditAudit code-to-schema fit — traces code paths to DB, identifies mismatches

Knowledge

SkillDescription
/learnIngest a document or repo into the shared knowledge store
/reflectPre-compact synthesis — capture session context before it evaporates
/indexOrganize the shared knowledge store; regenerate INDEX.md
/rememberPull relevant prior knowledge at session start
/publishDistribute knowledge to other surfaces (bots, docs sites)

Git

SkillDescription
/quick-commitStage all + commit with conventional commits format
/commit-push-prCommit, push, and open a PR

Data & Infrastructure

SkillDescription
/dbtdbt build, test, compile, full-refresh
/neon-queryRun SQL against Neon PostgreSQL via psql
/neon-infoNeon dashboard — connection status, tables, branches
/neon-branchCreate and manage Neon database branches

Platform Operations

SkillDescription
/modal-statusModal dashboard — apps, containers, secrets, GPU usage
/modal-deployDeploy Modal apps with health verification
/modal-logsView and stream Modal app logs
/railway-statusRailway dashboard — services, deployments, environments
/railway-deployDeploy to Railway with build monitoring and rollback
/railway-logsFilter Railway logs by level, time, patterns
/vercel-statusVercel dashboard — deployments, domains, env vars
/vercel-deployDeploy to Vercel with health verification and rollback
/vercel-logsFilter Vercel deployment logs
/investigate-appPlatform-agnostic production debugging

Utilities

SkillDescription
/notesPersistent notes across sessions
/find-skillsDiscover and install new agent skills
/heistRaid a GitHub repo for skills, config patterns, and novel approaches
/init-projectSet up CLAUDE.md, CHANGELOG.md, and .claude/ for a new or existing project
/file-github-issueFile a GitHub issue from a screenshot + short description
/github-activity-reportGitHub activity stats (PRs, commits, contributors) over a time window

Daily Workflow

The skills form a pipeline:

Morning
───────────────────────────────────────────────────
  /claudna:repo-health           → See what needs attention across all repos
  /claudna:session-resume        → Pick a repo, get a 30-second briefing

Planning (when starting new work)
───────────────────────────────────────────────────
  /claudna:tech-debt             → Find what's broken → generates phase docs
  /claudna:product-enhance       → Find what could be better → generates phase docs
  /claudna:product-vision        → Explore what the codebase could become

Building (the core loop)
───────────────────────────────────────────────────
  /claudna:implement-plan        → Point at a phase doc
                                   ├── Challenge round (first principles review)
                                   ├── Build & test
                                   ├── PR with deliverable audit
                                   └── Say "next" → auto-compacts → next phase

Reviewing
───────────────────────────────────────────────────
  /claudna:review-pr 42          → Structured review of any PR
  /claudna:review-changes        → Audit uncommitted work

End of Day
───────────────────────────────────────────────────
  /claudna:session-handoff       → Capture decisions, learnings, next steps
                                   → Writes handoff file for tomorrow's /claudna:session-resume

First time? Start with /claudna:repo-health to see your repos, then /claudna:session-resume in any project.

Agents

Specialized personas invoked by skills or directly:

"Use snowflake-analyst to find the largest tables"
"Use dbt-engineer to review this model"
"Use neon-analyst to check table sizes"
"Use railway-ops to diagnose the deployment failure"
"Use vercel-ops to investigate the 500 errors"
"Use modal-ops to check GPU utilization"

Safety

The marketplace install is managed by Claude Code: plugins live under ~/.claude/plugins/cache/Claudfather/claudna/<version>/. Updates land in a new version directory; old versions are kept for ~7 days before auto-cleanup.

Never touched by claudna:

  • ~/.claude/settings.json — user-managed. claudna never writes here. Recommended snippets are documented in SETUP_GUIDE for you to merge manually.
  • ~/.claude/notes/ — personal data (lessons, decisions, patterns).

Project Template

project-template/ provides a starter .claude/ directory for individual projects. It's not shipped via the plugin — copy it manually or use /claudna:init-project:

FilePurpose
CLAUDE.mdProject instructions with workflow principles
.claude/settings.jsonProject-specific permissions
.claude/lessons.mdProject-specific lessons from corrections

Contributing

See CONTRIBUTING.md for the full guide. Writing a new skill? Start with the Skill Authoring Guide.

Quick version: branch off main, make your change, run python3 scripts/validate-skills.py, update CHANGELOG.md, open a PR.

Documentation

FileDescription
PROJECT_MISSION.mdWhat clauDNA is, what it's becoming, and where it fits in the Claudfather ecosystem
CLAUDE.mdOperating rules for agents extending clauDNA
SKILL_CONTRACT.mdThe binding contract every skill must satisfy. Enforced in CI by scripts/validate-skills.py
SETUP_GUIDE.mdDeep-dive setup — settings.json bootstrap, headless/CI provisioning, Snowflake key-pair auth, troubleshooting
SKILL_AUTHORING_GUIDE.mdHow to write, test, and submit a skill — patterns, examples, anti-patterns
CONTRIBUTING.mdContribution workflow — bug reports, PRs, testing requirements
CLAUDE_MD_TEMPLATE.mdTemplate for project-level CLAUDE.md files (used by /claudna:init-project)
CHANGELOG.mdAll notable changes
Ver o original no GitHub

Bom para

Relacionados