UNPKG

major-ai-skills

Version:

Installable agentic skills / AI agent skills (SKILL.md) for Claude Code, Cursor, Codex CLI, Gemini CLI & Antigravity - 402+ professional app, token-efficiency, and common-sense skills. SEO/GEO ready.

114 lines (89 loc) 4.47 kB
--- name: reuse-saved-prompt-templates description: "Create parameterized prompt templates for recurring work and keep their assumptions and inputs explicit." category: common-sense risk: safe source: self source_type: self date_added: "2026-08-26" tags: ["prompt-library", "templates", "snippets", "textexpander", "productivity", "prompt-engineering"] tools: ["claude", "cursor", "gemini", "codex", "chatgpt"] --- # Reuse Saved Prompt Templates (Prompt Library Architecture) (AI Skill) ## Overview Re-typing complex prompt instructions from scratch every morning (*"Review this code, make sure you check for null values, do not write preambles..."*) is inefficient and leads to forgotten constraints and inconsistent outputs. The **Prompt Library Architecture** stores your highest-performing, battle-tested prompt recipes as parameterized templates in text expansion tools (Raycast, Alfred, TextExpander, or Obsidian), enabling you to trigger master-grade prompts with a 3-letter shortcut. --- ## Ad-Hoc Typing vs. Parameterized Snippet Library ``` ┌─────────────────────────────────────────────────────────────┐ Prompt Library Workflow Ad-Hoc Typing (Every Day): 2 minutes spent typing instructions from memory Forgets 2 constraints $\rightarrow$ mediocre response Parameterized Snippet (Type `;creview` or `;exec`): 1-second keyboard shortcut Injects 100% complete battle-tested constraints 10/10 Output on the First Try Every Single Time └─────────────────────────────────────────────────────────────┘ ``` --- ## The 5 Essential Core Templates ### 1. The Code Hardening Reviewer (Shortcut: `;creview`) ```markdown Act as a Principal Staff Engineer. Review this code for production readiness: <code_block> {clipboard} </code_block> Audit for: 1) Memory/concurrency leaks, 2) Missing null guards, 3) Deprecated APIs, 4) Edge-case exceptions. Provide the hardened production version with full type hints. ``` --- ### 2. The Meeting Action Parser (Shortcut: `;notes`) ```markdown Parse these raw meeting scribbles into: 1. **Key Decisions Made** (Max 3 bullets) 2. **Action Items Table**: | Task | Owner | Deadline | Priority | 3. **Open Blockers** Raw Notes: {clipboard} ``` --- ### 3. The Executive BLUF Digest (Shortcut: `;bluf`) ```markdown Read the attached text. Provide a 2-tier summary: 1. **Executive BLUF (Max 2 sentences)**: The direct decision and financial/timeline impact. 2. **3 Key Supporting Takeaways** (Bulleted with bold headers). Text: {clipboard} ``` --- ### 4. The Human Voice Copy Polish (Shortcut: `;polish`) ```markdown Refactor this draft to sound warm, confident, and conversational: - Ban all AI clichés: "delve", "tapestry", "crucial", "testament", "beacon". - Vary sentence rhythm (mix short and compound sentences). - Start immediately with the hook; zero preamble fluff. Draft: {clipboard} ``` --- ### 5. The TSV Data Extractor (Shortcut: `;table`) ```markdown Extract this unstructured text into a clean TSV code block ready for 1-click Excel paste. Columns: Date (YYYY-MM-DD), Entity/Person, Category, Amount (Number), Status. Data: {clipboard} ``` --- ## Implementation Setup Guide | Tool | Recommended Setup | | :--- | :--- | | **Raycast (Mac)** | Extensions $\rightarrow$ Snippets $\rightarrow$ Create Snippet with `{clipboard}` trigger. | | **TextExpander / Espanso (Cross-Platform)**| Define YAML snippets mapped to shortcut triggers like `;notes` or `;creview`. | | **Obsidian / Notion** | Create a `Prompts/` database categorized by Engineering, Writing, and Finance. | --- ## Summary Best Practice > **"Whenever an AI gives you a 10/10 response, don't just close the tab - extract the prompt, replace the variables with `{clipboard}`, and save it to your snippet library."**