{
  "markdown": "# Valantic Claude Code Marketplace\n\nCompany-internal Claude Code marketplace providing branded tools, workflows, and coding standards for Valantic developers.\n\n## Installation\n\n```bash\n# Add the Valantic marketplace\n/plugin marketplace add valantic-CEC-Deutschland-GmbH/claude-code-marketplace\n\n# Browse available plugins\n/plugins\n\n# Install individual plugins\n/plugin install valantic-brand-guidelines@valantic-claude-code-marketplace\n/plugin install valantic-documents@valantic-claude-code-marketplace\n/plugin install valantic-dev-workflow@valantic-claude-code-marketplace\n/plugin install valantic-coding-standards@valantic-claude-code-marketplace\n/plugin install valantic-project-setup@valantic-claude-code-marketplace\n/plugin install valantic-ai-permission-evaluator@valantic-claude-code-marketplace\n/plugin install valantic-fast-tools@valantic-claude-code-marketplace\n```\n\n## Available Plugins\n\n| Plugin | Type | Description |\n|--------|------|-------------|\n| `valantic-brand-guidelines` | Skill | Liquid Brand Design system (Routes A/B/C, colors, typography) |\n| `valantic-documents` | Skill | Branded PPTX/PDF/DOCX generation with Valantic templates |\n| `valantic-dev-workflow` | Commands + Hooks | Deployment, review, onboarding commands + permission hooks |\n| `valantic-coding-standards` | Skill + Hooks | Enforce company coding conventions automatically |\n| `valantic-project-setup` | Skill + Commands | Bootstrap new projects with Valantic templates (PHP, React, Spring) |\n| `valantic-ai-permission-evaluator` | Hooks | LLM-powered tool safety evaluation using GPT-4.1-nano |\n| `valantic-fast-tools` | Hooks + Agents | Force all agents to use fast CLI tools (rg, fd, bat) for faster context building |\n\n## Plugin Details\n\n### valantic-brand-guidelines\n\nApplies Valantic's official **Liquid Brand Design** system. Includes:\n- Visual Routes A (Restrained), B (Standard), C (Dynamic)\n- Official color palette (Coral `#FF5757`, Purple Heart `#4C26B7`, Peach `#FFCCAA`)\n- Typography system (Calibre/Segoe UI)\n- CSS variables and design tokens\n\n### valantic-documents\n\nGenerate branded documents:\n- **PPTX**: Presentations with Valantic slide layouts and styling\n- **PDF**: Reports and whitepapers with brand colors and typography\n- **DOCX**: Business documents with proper formatting\n\n### valantic-dev-workflow\n\nDevelopment workflow commands:\n- `/deploy` — Guided deployment with environment checks and rollback support\n- `/review` — Structured code review against Valantic standards\n- `/onboard` — New developer onboarding checklist\n\nIncludes a permission evaluator hook that enforces safe tool usage policies.\n\n### valantic-coding-standards\n\nAutomatic enforcement of Valantic coding conventions:\n- PHP: PSR-12, strict types, Spryker/Symfony patterns\n- TypeScript/React: functional components, strict mode, naming conventions\n- Java/Spring: standard patterns, proper annotations\n- General: error handling, logging, security practices\n\nPre-tool-use hook checks edits against standards.\n\n### valantic-project-setup\n\nBootstrap new projects:\n- `/new-project` — Interactive project scaffolding\n- PHP (Spryker/Symfony) template\n- React (Next.js/Vite) template\n- Spring Boot (Kotlin/Java) template\n\n### valantic-ai-permission-evaluator\n\nLLM-powered pre-tool-use hook that evaluates **every** Claude Code tool call for safety using OpenAI's GPT-4.1-nano model. Unlike the regex-based evaluator in `valantic-dev-workflow`, this uses an LLM to contextually assess whether a tool operation is safe.\n\n**Prerequisites:**\n- Node.js and npm installed\n- An OpenAI API key set as `OPENAI_API_KEY` environment variable\n\n**Setup:** Dependencies are installed automatically on first session start via a `SessionStart` hook. No manual `npm install` required.\n\n**How it works:**\n- Internal tools (TaskUpdate, TaskCreate, etc.) are auto-approved as always safe\n- All other tool calls are sent to GPT-4.1-nano for real-time safety evaluation\n- Safe operations are auto-approved; flagged operations prompt the user for confirmation\n- On evaluation failure, falls back to asking the user (never silently allows)\n\n**Comparison with regex-based evaluator:**\n\n| | Regex (`valantic-dev-workflow`) | AI (`valantic-ai-permission-evaluator`) |\n|---|---|---|\n| Scope | Bash commands only | All tool calls |\n| Approach | Pattern matching | LLM contextual evaluation |\n| Dependencies | Python stdlib only | Node.js + `openai` npm package |\n| External API | None | OpenAI API (`OPENAI_API_KEY` required) |\n| Latency | ~instant | ~200-500ms per evaluation |\n\n### valantic-fast-tools\n\nForces every Claude Code agent (including the built-in exploration agents) to use modern, fast CLI tools instead of `grep`/`find`/`cat` - searches that build agent context run 5-50x faster with ripgrep alone. A `PreToolUse` hook blocks slow commands with a corrective message so the agent retries with the fast tool, and three CLAUDE.md-aware exploration agents (`code-explorer`, `code-architect`, `code-reviewer`) are registered automatically.\n\n**System dependencies (install these or the hook stays passive):**\n\n```bash\n# Arch\nsudo pacman -S ripgrep fd bat tree jq\n# Debian / Ubuntu (symlink fdfind to fd)\nsudo apt install ripgrep fd-find bat tree jq\n# macOS\nbrew install ripgrep fd bat tree jq\n```\n\nThe hook is binary-aware: each rule only activates when the fast alternative is actually installed, and `jq` is required for the hook itself. Two optional one-line settings (`USE_BUILTIN_RIPGREP=0`, deny built-in `Explore`/`Plan` agents) unlock the full effect - see the [plugin README](plugins/valantic-fast-tools/README.md) for what is automated vs. manual.\n\n## GitHub Agents\n\nReusable GitHub Actions workflows powered by Claude Code CLI that run autonomously in CI. This collection is centralized here so every consuming repository only needs a thin caller workflow.\n\nSee **[github-agents.md](github-agents.md)** for the full catalog, setup instructions, and design rationale.\n\n## TweakCC UI Configuration\n\nThis repository includes a [`.tweakcc/config.json`](.tweakcc/config.json) that customizes the Claude Code terminal UI. It is managed by [TweakCC](https://github.com/Piebald-AI/tweakcc), a tool for theming and personalizing the Claude Code CLI experience.\n\n**What's configured:**\n- **8 color themes** — Dark, Light, ANSI-only, Colorblind-friendly, Monochrome, and Sweet\n- **Custom thinking verbs** — Replaces the default spinner text with personality (hmm, oof, yeet, segfault, ...)\n- **Braille spinner animation** — Progressive dot-fill pattern with reverse mirror\n- **Styled user messages** — Custom borders, colors, and formatting\n- **Misc tweaks** — Expanded thinking blocks, conversation titles, swarm mode, session memory, and more\n\n**Install TweakCC:**\n\n```bash\nnpm install -g tweakcc\n```\n\nAfter installation, TweakCC picks up the `.tweakcc/config.json` from the repo root and applies the configured themes and settings to your Claude Code session. See the [TweakCC documentation](https://github.com/Piebald-AI/tweakcc) for all available configuration options.\n\n## Security\n\n- **No secrets**: API keys, tokens, and passwords must stay in local environment variables\n- **Private repo**: Developers need GitHub access via SSH or HTTPS\n- **Hook scripts**: Use Python standard library only (no pip dependencies)\n- **MCP configs**: Use `${ENV_VAR}` patterns for credentials\n\n## Contributing\n\nSee [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines on adding or updating plugins.\n",
  "bytes": 7444,
  "sha": "b7a846c99e3eba9c48affb38aa729c5fe95e199731488bba16e91cdce275d403",
  "repo_slug": "valantic-cec-deutschland-gmbh/claude-code-marketplace",
  "fonte": "repo",
  "truncated": false,
  "api": "https://api.agentalog.com/api/listings/skl_valantic_cec_deutschland_gmbh_claude_cod_e73bf721/readme"
}