--- title: I built a prompt library that cuts Claude Code token... description: An open-sourced prompt library for Claude Code that claims 70-90% token usage reductions through reusable prompts, custom slash commands, subagents, and MCP... canonical: https://daily.dev/posts/i-built-a-prompt-library-that-cuts-claude-code-token-usage-by-70-90-here-s-how-6rjfhoud2 twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: I built a prompt library that cuts Claude Code token usage by 70-90% — Here's How | daily.dev og:description: An open-sourced prompt library for Claude Code that claims 70-90% token usage reductions through reusable prompts, custom slash commands, subagents, and MCP... og:url: https://daily.dev/posts/i-built-a-prompt-library-that-cuts-claude-code-token-usage-by-70-90-here-s-how-6rjfhoud2 og:image: https://api.daily.dev/og/posts/6rjFhouD2.png og:image:alt: I built a prompt library that cuts Claude Code token usage by 70-90% — Here's How og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # I built a prompt library that cuts Claude Code token usage by 70-90% — Here's How **[Nikola Katsarov](https://daily.dev/sources/83l3nmc1k)** · [@katsarov](https://daily.dev/katsarov) · 3 min read · 0 upvotes · 0 comments ## Summary An open-sourced prompt library for Claude Code that claims 70-90% token usage reductions through reusable prompts, custom slash commands, subagents, and MCP integrations. The system involves a one-time global setup (2-3 hours) plus per-project activation (10-15 min). Key techniques include a symbol-first protocol using Serena for code navigation, specialized subagents with controlled tool access, security hardening against prompt injection, and cross-platform guides for web, mobile, and desktop development. Real-world examples show savings from 73-93% on tasks like bug fixes, feature development, and module creation. ## Content If you've been using Claude Code, you know the pain: token usage spiraling out of control, costs adding up, and sessions that feel wasteful. I spent months refining a system of reusable prompts, skills, and agents that consistently **cut token usage by 70-90%** across all my projects. I've open-sourced everything as a prompt library: [github.com/escapeboy/ai-prompts](https://github.com/escapeboy/ai-prompts) ## What is it? A collection of project-agnostic guides and executable agents for setting up Claude Code with maximum efficiency. It includes: - **Global optimization setup** — one-time configuration (~2-3 hours) that applies to ALL your projects - **Project activation** — 10-15 min per project to enable project-specific memories and conventions - **Custom skills (slash commands)** — `/optimize`, `/context`, `/cache-inspector`, `/init-project`, and more - **Subagents** — specialized AI agents (code reviewer, debugger, test generator) with controlled tool access - **MCP integrations** — Laravel Boost, WebMCP, Chrome DevTools, and more - **Security hardening** — protect your workflows from prompt injection and MCP supply chain attacks - **Mobile & Desktop development** — platform-specific guides for iOS, Android, React Native, Flutter, macOS, Tauri, and Electron ## Real-world results Here's what I'm seeing across actual projects: - **Simple bug fix:** 22,000 → 1,600 tokens (**93% savings**) - **New feature:** 31,000 → 8,500 tokens (**73% savings**) - **Module creation:** 85,000 → 18,000 tokens (**79% savings**) That translates to **$700-$2,000+ saved per year** depending on your usage volume. ## How to get started **Option 1: Automated setup (recommended)** Navigate to `~/.claude`, copy the contents of `01-global-optimization/setup-agent.md`, and paste it into a Claude Code conversation. The agent handles everything automatically. **Option 2: Manual setup** Follow the step-by-step guide in `01-global-optimization/guide.md`, create files as instructed, then verify with the checklist. After global setup, activate any project in 10-15 minutes using the activation agent. ## Key features worth highlighting 🔧 **Symbol-first protocol** — 65-75% token savings by using Serena for code navigation instead of reading full files 🤖 **Agent Teams** — native peer-to-peer coordination on shared codebases (Claude Code v2.1.32+) 🛡️ **Security hardening** — MCP vetting checklists, prompt injection defense hooks, and production safety rules 📱 **Cross-platform** — guides for web, mobile (iOS/Android/RN/Flutter), and desktop (macOS/Tauri/Electron) 🎨 **UI/UX Pro skill** — 50+ UI styles, 21 color palettes, shadcn/ui integration with 40-60% token savings ## Who is this for? Anyone using Claude Code who wants to spend less on tokens while getting better results. Whether you're a solo developer or part of a team, the library is designed to be shared, forked, and adapted to your specific workflow. The setup is a one-time investment of 2-3 hours for global configuration, plus 10-15 minutes per project. Every future project benefits automatically. **Check it out:** [github.com/escapeboy/ai-prompts](https://github.com/escapeboy/ai-prompts) Would love to hear your feedback, suggestions, or how you're optimizing your own Claude Code workflows! ## Similar posts on daily.dev - [GitHub - drona23/claude-token-efficient: Universal CLAUDE.md - cut Claude output tokens by 63%. Drop-in. No code changes.](https://daily.dev/posts/github---drona23-claude-token-efficient-universal-claude-md---cut-claude-output-tokens-by-63-drop-cba3nnscu) · Hacker News · 3 upvotes · 0 comments - [Claude API Token Optimization](https://daily.dev/posts/claude-api-token-optimization-4sbqfmbsk) · SitePoint · 0 upvotes · 0 comments --- Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#prompt-engineering](https://daily.dev/tags/prompt-engineering), [#mcp](https://daily.dev/tags/mcp), [#claude-code](https://daily.dev/tags/claude-code) [View this post on daily.dev](https://daily.dev/posts/i-built-a-prompt-library-that-cuts-claude-code-token-usage-by-70-90-here-s-how-6rjfhoud2) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"DiscussionForumPosting","mainEntityOfPage":"https://daily.dev/posts/i-built-a-prompt-library-that-cuts-claude-code-token-usage-by-70-90-here-s-how-6rjfhoud2","headline":"I built a prompt library that cuts Claude Code token usage by 70-90% — Here's How","text":"An open-sourced prompt library for Claude Code that claims 70-90% token usage reductions through reusable prompts, custom slash commands, subagents, and MCP integrations. The system involves a one-time global setup (2-3 hours) plus per-project activation (10-15 min). Key techniques include a symbol-first protocol using Serena for code navigation, specialized subagents with controlled tool access, security hardening against prompt injection, and cross-platform guides for web, mobile, and desktop development. Real-world examples show savings from 73-93% on tasks like bug fixes, feature development, and module creation.","url":"https://daily.dev/posts/i-built-a-prompt-library-that-cuts-claude-code-token-usage-by-70-90-here-s-how-6rjfhoud2","datePublished":"2026-04-01T05:27:09.022Z","dateModified":"2026-04-01T05:27:43.657Z","author":{"@type":"Person","name":"Nikola Katsarov","url":"https://daily.dev/katsarov","image":"https://lh3.googleusercontent.com/a-/AOh14GgpvRFkJxQj1aEm7n-js-vcxmGsqzjrsM61bK5UKFE=s100","description":"Ex-Full-Stack Developer & AI Tooling Engineer","interactionStatistic":{"@type":"InteractionCounter","interactionType":{"@type":"EndorseAction"},"userInteractionCount":160}},"image":"https://media.daily.dev/image/upload/s--MnU13uld--/f_auto/v1775021229/posts/6rjFhouD2?_a=BAMAMiWQ0","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"isPartOf":{"@type":"WebPage","url":"https://daily.dev/sources/83l3nmc1k","name":"Nikola Katsarov"}} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Nikola Katsarov","item":"https://daily.dev/sources/83l3nmc1k"},{"@type":"ListItem","position":3,"name":"I built a prompt library that cuts Claude Code token usage by 70-90% — Here's How"}]} ```