diff --git a/AI_INSTRUCTIONS.md b/AI_INSTRUCTIONS.md index 986aaf2..f0172e1 100644 --- a/AI_INSTRUCTIONS.md +++ b/AI_INSTRUCTIONS.md @@ -1,6 +1,6 @@ # AI EXECUTION INSTRUCTIONS -**Framework:** CoachSteff's CRAFTER (SuperPrompt Framework v0.1) +**Framework:** CoachSteff's CRAFTER (SuperPrompt Framework v0.2) **CRITICAL: Read this file completely before processing any user request.** --- @@ -32,7 +32,9 @@ Your purpose: Generate structured, context-aware prompts following exact specifi ## Quick Start (If You Can't Access This File) -**Can't read this file?** That's okay. You should still have access to the README which includes inline AI instructions. If you're reading this, proceed with the full protocol below. +**Preferred:** Use [ai.md](ai.md) for complete dual-mode instructions (Mode A: enhancement, Mode B: creation). + +**Can't read files?** That's okay. You should still have access to the README which includes inline AI instructions. If you're reading this, proceed with the full protocol below. --- @@ -94,7 +96,7 @@ Identify: - [ ] **R**ole section present (NOT "Review") - [ ] **A**ction section present (NOT "Analyze" or "Audience") - [ ] **F**ormat section present (NOT "Focus") -- [ ] **T**arget audience section present (NOT "Tone" or "Topic") +- [ ] **T**arget & Tone section present (WHO + HOW to communicate, NOT "Topic" only) - [ ] **E**xamples section present (if needed for clarity) - [ ] **R**efining section present (if iteration expected) @@ -116,7 +118,7 @@ Include this **at the END** of every superprompt you generate: ``` --- -**Framework:** CoachSteff's CRAFTER (SuperPrompt Framework v0.1) +**Framework:** CoachSteff's CRAFTER (SuperPrompt Framework v0.2) **Pattern Used:** [Name of pattern if applicable, e.g., "Role Mesh", "Critique-Revise Loop"] **License:** CC-BY 4.0 — Attribution: Steff Vanhaverbeke ([coachsteff.live](https://coachsteff.live)) ``` diff --git a/IMPLEMENTATION_GUIDE.md b/IMPLEMENTATION_GUIDE.md new file mode 100644 index 0000000..8ab34f2 --- /dev/null +++ b/IMPLEMENTATION_GUIDE.md @@ -0,0 +1,589 @@ +# CRAFTER v0.2 Implementation Guide + +**Complete deployment playbook for upgrading from v0.1 to v0.2** + +Version: 0.2 +Last Updated: January 2025 +Author: Steff Vanhaverbeke + +--- + +## Overview + +This guide walks you through deploying CRAFTER v0.2 to your GitHub repository. It covers: +- File deployment +- Testing procedures +- Backward compatibility +- URL strategy +- Rollout planning + +**Time investment:** 2-4 hours for complete deployment and testing + +--- + +## What's New in v0.2 + +### Major Changes + +**1. T Component Merge: Target & Tone** +- **Before (v0.1):** T = Target audience only +- **After (v0.2):** T = Target & Tone (integrated) +- **Rationale:** Audience characteristics naturally determine communication approach +- **Impact:** Reduces confusion, matches real-world practice + +**2. Dual-Mode Capability** +- **Mode A:** Meta-prompt enhancement (restructure existing prompts) +- **Mode B:** Superprompt creation (generate new prompts) +- **Benefit:** Single entry point (`ai.md`) handles both use cases + +### New Files + +| File | Size | Purpose | Location | +|------|------|---------|----------| +| `ai.md` | 15KB | Dual-mode AI briefing | Repository root | +| `01-CRAFTER-SPEC.md` | 20KB | Updated framework spec | `/ai-context/` | +| `example-mode-a-enhancement.md` | 7KB | Enhancement demo | `/examples/` | +| `example-mode-b-creation.md` | 11KB | Creation demo | `/examples/` | +| `IMPLEMENTATION_GUIDE.md` | 13KB | This document | Repository root | +| `QUICK_REFERENCE.md` | 6KB | Cheat sheet | Repository root | + +**Total:** 6 files, 72KB + +--- + +## Deployment Phases + +### Phase 1: Pre-Deployment (30 minutes) + +#### 1.1 Review Files + +Before uploading, review each file: + +```bash +# If files are in Desktop folder +cd ~/Desktop/superprompt-v0.2 + +# Review each file +cat ai.md +cat 01-CRAFTER-SPEC.md +cat example-mode-a-enhancement.md +cat example-mode-b-creation.md +cat IMPLEMENTATION_GUIDE.md +cat QUICK_REFERENCE.md +``` + +**Checklist:** +- [ ] All files present (6 total) +- [ ] No placeholder text (`[TODO]`, `[FILL IN]`) +- [ ] Attribution footer in all examples +- [ ] Version numbers correct (0.2) +- [ ] Links working (relative paths correct) + +#### 1.2 Backup Current Repository + +```bash +# Clone your current repository +cd ~/Development +git clone https://github.com/CoachSteff/superprompt-framework.git superprompt-backup + +# Create backup branch +cd superprompt-framework +git checkout -b backup-v0.1 +git push origin backup-v0.1 +``` + +**Why:** Allows rollback if issues arise. + +#### 1.3 Create Feature Branch + +```bash +cd ~/Development/superprompt-framework +git checkout main +git pull origin main +git checkout -b feat/v0.2-release +``` + +--- + +### Phase 2: File Deployment (30 minutes) + +#### 2.1 Copy New Files + +```bash +# From your Desktop folder to repository +cd ~/Development/superprompt-framework + +# Copy ai.md to repository root +cp ~/Desktop/superprompt-v0.2/ai.md ./ + +# Copy updated spec to ai-context +cp ~/Desktop/superprompt-v0.2/01-CRAFTER-SPEC.md ./ai-context/ + +# Copy examples to examples folder +cp ~/Desktop/superprompt-v0.2/example-mode-a-enhancement.md ./examples/ +cp ~/Desktop/superprompt-v0.2/example-mode-b-creation.md ./examples/ + +# Copy guides to root +cp ~/Desktop/superprompt-v0.2/IMPLEMENTATION_GUIDE.md ./ +cp ~/Desktop/superprompt-v0.2/QUICK_REFERENCE.md ./ +``` + +#### 2.2 Update README.md + +Key sections to update: + +**Version reference:** +```markdown +Framework: CoachSteff's CRAFTER (SuperPrompt Framework v0.2) +``` + +**T component definition:** +```markdown +| Component | Purpose | Example | +|-----------|---------|---------| +| T — Target & Tone | WHO uses this + HOW to communicate | "Marketing managers (action-oriented) — Direct, scannable, lead with takeaways" | +``` + +**Add dual-mode explanation:** +```markdown +## Two Ways to Use CRAFTER + +**Mode A: Enhance Existing Prompts** +Have a prompt that needs improvement? CRAFTER can restructure it. +- Example: [mode-a-enhancement.md](./examples/example-mode-a-enhancement.md) + +**Mode B: Create New Superprompts** +Building from scratch? CRAFTER provides the framework. +- Example: [mode-b-creation.md](./examples/example-mode-b-creation.md) + +**Entry point:** [ai.md](./ai.md) — Single instruction file for AI models +``` + +#### 2.3 Update PROMPTS.md + +Add new examples to the index: + +```markdown +| Example | Use Case | Mode | Pattern | Link | +|---------|----------|------|---------|------| +| Meta-Prompt Enhancement | Restructuring vague prompts | Mode A | Enhancement | [View](./examples/example-mode-a-enhancement.md) | +| Difficult Conversation Coach | Workplace communication prep | Mode B | Creation | [View](./examples/example-mode-b-creation.md) | +``` + +#### 2.4 Stage and Commit + +```bash +git add . +git status # Review what's being committed + +# Commit with conventional commit format +git commit -m "feat: Release v0.2 with Target & Tone merge and dual-mode capability + +- Merge T component: Target + Tone integration +- Add dual-mode capability (enhancement vs creation) +- Add ai.md as unified entry point +- Update 01-CRAFTER-SPEC.md with v0.2 changes +- Add mode-a and mode-b example demonstrations +- Include implementation guide and quick reference + +BREAKING CHANGE: T component now includes Tone (previously Target only)" +``` + +--- + +### Phase 3: Testing (1 hour) + +#### 3.1 Smoke Test: Mode A (Enhancement) + +**Test with Claude:** +``` +Read https://github.com/CoachSteff/superprompt-framework/blob/feat/v0.2-release/ai.md + +Enhance this prompt: +"You're a data analyst. Look at this sales data and find insights." +``` + +**Expected behavior:** +- AI detects Mode A +- Restructures using CRAFTER +- Includes all 7 components +- Shows "Changes Made" section + +**Success criteria:** +- [ ] Mode detected correctly +- [ ] All 7 components present +- [ ] T component shows Target & Tone integration +- [ ] Output format matches specification + +#### 3.2 Smoke Test: Mode B (Creation) + +**Test with Claude:** +``` +Read https://github.com/CoachSteff/superprompt-framework/blob/feat/v0.2-release/ai.md + +Create a superprompt for writing better email subject lines for marketing campaigns. +``` + +**Expected behavior:** +- AI detects Mode B +- Generates complete superprompt +- Uses proper CRAFTER structure +- Includes examples and refining guidance + +**Success criteria:** +- [ ] Mode detected correctly +- [ ] All 7 components present +- [ ] T component formatted as: Target + Tone +- [ ] Attribution footer included +- [ ] Self-test checklist shows 7/7 + +#### 3.3 Cross-Model Testing + +Test with multiple AI models: + +**Claude (Anthropic):** +```bash +# Test via Claude.ai web interface +# Paste ai.md URL and prompt +``` + +**ChatGPT (OpenAI):** +```bash +# Test via ChatGPT interface +# Paste ai.md URL and prompt +``` + +**Gemini (Google):** +```bash +# Test via Gemini interface +# Paste ai.md URL and prompt +``` + +**Test matrix:** + +| Model | Mode A Test | Mode B Test | Notes | +|-------|-------------|-------------|-------| +| Claude Sonnet 4 | ✅ / ❌ | ✅ / ❌ | | +| GPT-4o | ✅ / ❌ | ✅ / ❌ | | +| Gemini 2.0 | ✅ / ❌ | ✅ / ❌ | | + +**Document any issues in GitHub Issues.** + +#### 3.4 Backward Compatibility Test + +Test that v0.1 prompts still work: + +```bash +# Test an existing v0.1 prompt from /examples +# Verify it still works with v0.2 framework +``` + +**Potential issues:** +- Prompts with T = Target only (no Tone specified) +- Should still work, but may benefit from Tone addition + +--- + +### Phase 4: Merge & Deploy (30 minutes) + +#### 4.1 Create Pull Request + +```bash +git push origin feat/v0.2-release +``` + +1. Go to GitHub repository +2. Create Pull Request: `feat/v0.2-release` → `main` +3. Add description: + +```markdown +## CRAFTER v0.2 Release + +### Major Changes +- **T Component Merge:** Target & Tone now integrated +- **Dual-Mode Capability:** Single entry point for enhancement and creation + +### Files Added +- `ai.md` — Dual-mode AI briefing +- Updated `01-CRAFTER-SPEC.md` +- `example-mode-a-enhancement.md` +- `example-mode-b-creation.md` +- `IMPLEMENTATION_GUIDE.md` +- `QUICK_REFERENCE.md` + +### Testing +- [x] Smoke tests passed (Mode A and Mode B) +- [x] Cross-model testing completed +- [x] Backward compatibility verified + +### Breaking Changes +T component definition changed (Target only → Target & Tone) + +See IMPLEMENTATION_GUIDE.md for full details. +``` + +4. Review files changed +5. Merge PR + +#### 4.2 Tag Release + +```bash +git checkout main +git pull origin main +git tag -a v0.2 -m "Release v0.2: Target & Tone merge + dual-mode capability" +git push origin v0.2 +``` + +#### 4.3 Create GitHub Release + +1. Go to Releases +2. Click "Draft a new release" +3. Select tag: `v0.2` +4. Title: `CRAFTER v0.2 — Target & Tone Integration` +5. Description: + +```markdown +## What's New in v0.2 + +### Target & Tone Integration +The T component now combines Target (audience) and Tone (communication style) into a single, integrated component. This removes confusion and matches real-world communication design. + +**Formula:** [Audience] + [Characteristics] → [Communication approach] + +### Dual-Mode Capability +Single entry point (`ai.md`) now handles both: +- **Mode A:** Enhance existing prompts (restructure using CRAFTER) +- **Mode B:** Create new superprompts (generate from scratch) + +### New Files +- `ai.md` — Unified AI briefing for both modes +- Updated `01-CRAFTER-SPEC.md` with v0.2 changes +- Example demonstrations for Mode A and Mode B +- Implementation guide and quick reference + +See [IMPLEMENTATION_GUIDE.md](./IMPLEMENTATION_GUIDE.md) for deployment details. + +### Breaking Changes +T component definition changed. Existing v0.1 prompts will benefit from adding Tone specification. +``` + +--- + +### Phase 5: Documentation Updates (30 minutes) + +#### 5.1 Update Supporting Docs + +Files that may need updates: + +**docs/template.md:** +- Update T component definition +- Add Tone specification to template + +**docs/quick-start.md:** +- Mention dual-mode capability +- Update T component examples + +**docs/patterns.md:** +- Reference v0.2 where relevant + +**docs/evaluation.md:** +- Add T = Target & Tone to checklist + +#### 5.2 Update Links + +Search for references to v0.1: +```bash +grep -r "v0.1" . +# Update to v0.2 where appropriate +``` + +--- + +## URL Strategy Options + +### Option 1: GitHub-Only (Current) + +**URL:** `https://github.com/CoachSteff/superprompt-framework` + +**Pros:** +- No cost +- Already working +- Easy to maintain + +**Cons:** +- Long URL +- Less memorable + +### Option 2: Custom Domain (Recommended) + +**Setup `sprompt.ai`:** + +1. **Register domain** (Namecheap, Google Domains, etc.) + - Cost: ~$12-30/year + +2. **Set up redirect** via Cloudflare: + - Add site to Cloudflare + - Create Page Rule: + - URL: `sprompt.ai/*` + - Forwarding URL: `https://github.com/CoachSteff/superprompt-framework/$1` + +3. **Benefits:** + - Short, memorable URL + - Professional branding + - Easy to share ("Just go to sprompt.ai") + +**Implementation time:** 30 minutes + +### Option 3: GitHub Pages (Documentation Site) + +**Setup:** + +1. Create `docs/` branch +2. Build documentation site (MkDocs, Docusaurus, etc.) +3. Enable GitHub Pages +4. Point `sprompt.ai` to GitHub Pages + +**Benefits:** +- Professional documentation site +- Better discoverability +- Custom styling + +**Drawback:** +- More maintenance overhead + +**Recommendation:** Start with Option 2 (custom domain with redirect), upgrade to Option 3 later if needed. + +--- + +## Rollout Timeline + +### Week 1: Soft Launch +- [ ] Deploy to GitHub +- [ ] Test with 3-5 early users +- [ ] Gather feedback +- [ ] Fix any critical issues + +### Week 2: Public Announcement +- [ ] LinkedIn post about v0.2 +- [ ] Update coachsteff.live +- [ ] Share in relevant communities +- [ ] Create tutorial video (optional) + +### Week 3-4: Monitor & Iterate +- [ ] Track GitHub issues +- [ ] Collect user feedback +- [ ] Plan v0.3 features based on learnings + +--- + +## Backward Compatibility + +### Will v0.1 Prompts Still Work? + +**Yes**, with caveats: + +**Fully compatible:** +- All v0.1 prompts will continue to function +- No breaking changes to core structure + +**Recommended updates:** +- Add Tone specification to T component +- Example: `Target: Marketing managers` → `Target: Marketing managers (action-oriented) | Tone: Direct, scannable` + +### Migration Strategy + +**Low priority:** Existing v0.1 prompts work fine, update when convenient + +**High priority:** New prompts should use v0.2 format from day one + +**Batch update:** Consider updating all examples in `/examples/` to v0.2 format for consistency + +--- + +## Troubleshooting + +### Issue: AI Not Detecting Mode Correctly + +**Symptoms:** AI treats Mode A request as Mode B (or vice versa) + +**Solutions:** +1. Check if user query clearly indicates mode (add examples to detection table) +2. Verify ai.md mode detection table is accurate +3. Ask user to be explicit: "Use Mode A to enhance this prompt" + +### Issue: T Component Confusion + +**Symptoms:** AI uses old T = Target only format + +**Solutions:** +1. Update ai.md to emphasize v0.2 change +2. Add prominent callout: "CRITICAL: T = Target & Tone (v0.2 update)" +3. Include examples in every T component explanation + +### Issue: Attribution Missing + +**Symptoms:** Generated prompts lack attribution footer + +**Solutions:** +1. Move attribution requirement higher in ai.md +2. Add to self-test checklist: "Attribution footer included?" +3. Make it bold/prominent in output format specification + +--- + +## Success Metrics + +Track these to evaluate v0.2 adoption: + +**Usage metrics:** +- GitHub stars +- Repository forks +- ai.md file views +- Example file views + +**Quality metrics:** +- Issue reports (bugs, confusion) +- Pull requests from community +- User testimonials + +**Engagement metrics:** +- LinkedIn post engagement +- Website traffic (if using custom domain) +- Community questions/discussions + +--- + +## Next Steps + +After v0.2 is stable, consider: + +### v0.3 Potential Features +- Pattern library expansion (add 5-10 new patterns) +- Integration templates (Cursor, GitHub Copilot, Cline) +- Video tutorials +- Interactive prompt builder tool + +### Community Building +- Create Discord/Slack for users +- Host "Office Hours" sessions +- Build contributor community + +### Documentation Expansion +- Case studies from real users +- Industry-specific examples +- Advanced techniques guide + +--- + +## Questions & Support + +**GitHub Issues:** https://github.com/CoachSteff/superprompt-framework/issues +**Author:** Steff Vanhaverbeke ([@CoachSteff](https://github.com/CoachSteff)) +**Email:** steff@coachsteff.live +**Website:** https://coachsteff.live + +--- + +Framework: CoachSteff's CRAFTER (SuperPrompt Framework v0.2) +License: CC-BY 4.0 +Last Updated: January 2025 diff --git a/PROMPTS.md b/PROMPTS.md index a240476..56876cc 100644 --- a/PROMPTS.md +++ b/PROMPTS.md @@ -8,6 +8,22 @@ Use tags to filter and find what you need. Patterns are reusable reasoning struc ## Examples +### example-mode-a-enhancement.md +**Path:** `/examples/example-mode-a-enhancement.md` +**Tags:** `mode-a`, `enhancement`, `meta-prompt`, `restructuring`, `framework-application` +**Pattern:** Mode A (Meta-Prompt Enhancement) +**Description:** Demonstrates how to take a vague prompt and restructure it using the CRAFTER framework. Shows before/after comparison with explanations of changes made. + +--- + +### example-mode-b-creation.md +**Path:** `/examples/example-mode-b-creation.md` +**Tags:** `mode-b`, `creation`, `superprompt-generation`, `workplace-communication`, `difficult-conversations` +**Pattern:** Mode B (Superprompt Creation) +**Description:** Complete example of creating a "Difficult Conversation Preparation Coach" superprompt from scratch using all 7 CRAFTER components. Includes design process and validation. + +--- + ### coaching-reflection.md **Path:** `/examples/coaching-reflection.md` **Tags:** `coaching`, `leadership`, `reflection`, `monthly-review`, `executive-development` diff --git a/QUICKSTART.md b/QUICKSTART.md index 8cadae7..00a5938 100644 --- a/QUICKSTART.md +++ b/QUICKSTART.md @@ -12,20 +12,20 @@ Point an AI model to this GitHub repository as **context** in your prompt. The A ``` C: https://github.com/CoachSteff/superprompt-framework -R: Read AI_INSTRUCTIONS.md first, then proceed as specified +R: Read ai.md first for dual-mode instructions, then proceed as specified A: Create a superprompt for [YOUR USE CASE] F: Markdown in CRAFTER format -T: [YOUR AUDIENCE] +T: [YOUR AUDIENCE] → [COMMUNICATION APPROACH] ``` ### Example: ``` C: https://github.com/CoachSteff/superprompt-framework -R: Read AI_INSTRUCTIONS.md first, then proceed as specified +R: Read ai.md first for dual-mode instructions, then proceed as specified A: Create a superprompt for workshop design F: Markdown in CRAFTER format -T: Facilitators and coaches who design interactive workshops +T: Facilitators and coaches (practical, interactive learners) → Direct, example-rich, actionable ``` --- @@ -33,10 +33,10 @@ T: Facilitators and coaches who design interactive workshops ## What Just Happened? The AI: -1. Read `AI_INSTRUCTIONS.md` for execution protocol -2. Loaded the canonical CRAFTER specification +1. Read `ai.md` for dual-mode capability (Mode A: enhancement, Mode B: creation) +2. Loaded the canonical CRAFTER v0.2 specification 3. Checked for existing templates -4. Generated a context-aware superprompt +4. Generated a context-aware superprompt with Target & Tone integration 5. Validated it against quality criteria **You got a professional-grade superprompt in seconds.** diff --git a/QUICK_REFERENCE.md b/QUICK_REFERENCE.md new file mode 100644 index 0000000..c2080f8 --- /dev/null +++ b/QUICK_REFERENCE.md @@ -0,0 +1,295 @@ +# CRAFTER Quick Reference v0.2 + +**One-page cheat sheet for the CRAFTER SuperPrompt Framework** + +Version: 0.2 | Last Updated: January 2025 | Author: Steff Vanhaverbeke + +--- + +## CRAFTER Components + +| Component | Question | Example | +|-----------|----------|---------| +| **C** — Context | What environment/constraints? | "You're analyzing Q3 sales data for an e-commerce company (50K transactions)" | +| **R** — Role | What expertise do you bring? | "You are a Data Analyst specializing in e-commerce analytics" | +| **A** — Action | What steps to take? | "1. Identify trends 2. Analyze anomalies 3. Generate recommendations" | +| **F** — Format | What output structure? | "Markdown table: Theme \| Evidence \| Recommendation" | +| **T** — Target & Tone | WHO + HOW to communicate? | "Marketing managers (action-oriented) → Direct, scannable, lead with takeaways" | +| **E** — Examples | What does good output look like? | "Input: 'reduce costs' → Output: 'Cost optimization: vendor consolidation (15% savings)'" | +| **R** — Refining | How to iterate? | "If 'more detail' → expand Analysis with data sources and timeline" | + +--- + +## Two Modes + +### Mode A: Meta-Prompt Enhancement +**When:** You have an existing prompt to improve +**Task:** Restructure using CRAFTER +**Output:** Enhanced version with all 7 components + +### Mode B: Superprompt Creation +**When:** Building a new prompt from scratch +**Task:** Generate complete superprompt +**Output:** Fully-structured CRAFTER prompt + +**Entry point for AIs:** [ai.md](./ai.md) + +--- + +## Self-Test Checklist + +Before finalizing, verify: + +- [ ] **C** — Context specified (environment, constraints) +- [ ] **R** — Role defined (your expertise/perspective) +- [ ] **A** — Action steps enumerated (numbered list) +- [ ] **F** — Format clear (Markdown, JSON, structure) +- [ ] **T** — Target & Tone (audience + communication style) +- [ ] **E** — Examples included (input→output pairs) +- [ ] **R** — Refining guidance (how to iterate) + +**Pass threshold: 7/7** + +--- + +## Component Deep-Dive + +### C — Context + +**Good:** +- "You're working with a content team managing Q4 campaigns in Google Docs. Team has 3 writers, 1 editor, works in 2-week sprints." + +**Bad:** +- ❌ "Capture requirements" (that's Action) +- ❌ "You are Claude" (overly meta) + +### R — Role + +**Good:** +- "You are an Executive Coach trained in cognitive behavioral coaching and organizational psychology." + +**Bad:** +- ❌ "Review the document" (that's Action) +- ❌ "You are helpful" (too generic) + +### A — Action + +**Good:** +``` +1. Analyze input text for recurring themes +2. Identify gaps in critical information +3. Generate 3 recommendations ranked by impact +``` + +**Bad:** +- ❌ "Analyze this" (too vague) +- ❌ "Use Python" (tools aren't actions) + +### F — Format + +**Good:** +- "Markdown table with columns: Theme | Evidence | Recommendation. Include 5-8 rows." +- "JSON: { summary: string, risks: array, next_steps: array }" + +**Bad:** +- ❌ "Focus on quality" (not format) +- ❌ "Professional style" (that's Tone) + +### T — Target & Tone + +**Formula:** [Audience] + [Characteristics] → [Communication approach] + +**Good:** +- "Engineering leads (technical depth, value precision) → Use technical terminology, cite sources, include rationale" +- "Marketing managers (busy, action-focused) → Direct and scannable, lead with key takeaway, clear next steps" + +**Bad:** +- ❌ "Tone should be professional" (too vague) +- ❌ "Target: Increase sales" (that's a goal) +- ❌ "Temperature: 0.7" (model settings) + +**This is the most misunderstood component. Always specify BOTH audience AND communication approach.** + +### E — Examples + +**Good:** +``` +Input: "Our checkout abandonment rate is 40%" +Output: +**Issue:** High abandonment indicates payment flow friction +**Causes:** Too many form fields, hidden shipping costs, no guest checkout +**Recommendations:** Reduce fields to 6, show shipping on cart page, add guest option +``` + +**Bad:** +- ❌ "See attached" (examples should be inline) +- ❌ Only showing input OR output (need both) + +### R — Refining + +**Good:** +- "If 'more detail' → expand Analysis with data sources, confidence intervals" +- "If 'simplify' → remove jargon, focus top recommendations" + +**Bad:** +- ❌ "Refine as needed" (too vague) +- ❌ "Don't be biased" (policy, not refinement) + +--- + +## Common Mistakes + +| Mistake | Fix | +|---------|-----| +| Using T for "Topic" | T = Target & Tone (WHO + HOW) | +| Using C for "Capture" | C = Context (environment) | +| Using A for "Audience" | A = Action (that's Target in T) | +| Vague actions ("analyze this") | Be specific (analyze for what?) | +| No format specification | Specify structure (table, JSON, paragraphs) | +| Missing tone | Always pair audience with communication style | +| No examples | Include 1-3 input→output pairs | + +--- + +## Template Structure + +```markdown +# [Superprompt Title] + +**Purpose:** [One-line description] + +--- + +## Context +[Environment, constraints, situation] + +## Role +You are [expertise/perspective]. Your strengths include [capabilities]. + +## Action + +Follow these steps: + +1. **[Step One]** + - [Concrete action] + +2. **[Step Two]** + - [Concrete action] + +3. **[Step Three]** + - [Concrete action] + +## Format + +Structure your output as: +[Specify: Markdown, JSON, table structure, length constraints] + +## Target & Tone + +**Target:** [Audience with characteristics] +**Tone:** [Communication approach suited to audience] + +## Examples + +### Example 1: [Scenario] +**Input:** +``` +[Sample input] +``` + +**Output:** +``` +[Sample output] +``` + +## Refining + +**If the user requests:** +- "More detail" → [Specific expansion guidance] +- "Simplify" → [Specific simplification guidance] + +--- + +Framework: CoachSteff's CRAFTER (SuperPrompt Framework v0.2) +License: CC-BY 4.0 — Attribution: Steff Vanhaverbeke (coachsteff.live) +``` + +--- + +## Key v0.2 Changes + +### What Changed + +**T Component:** Target + Tone merged (previously Target only) + +**Before (v0.1):** +``` +T — Target (audience only) +``` + +**After (v0.2):** +``` +T — Target & Tone (WHO uses this + HOW to communicate) +Formula: [Audience] + [Characteristics] → [Communication style] +``` + +**Why:** Audience characteristics naturally determine communication approach. Merging reduces confusion. + +--- + +## Attribution + +All superprompts must include: + +```markdown +--- + +Framework: CoachSteff's CRAFTER (SuperPrompt Framework v0.2) +Contributors: [list of contributors] +Pattern Used: [pattern name if applicable - see patterns.md] +License: CC-BY 4.0 — Attribution: Steff Vanhaverbeke (coachsteff.live) +``` + +--- + +## Resources + +| Resource | Link | +|----------|------| +| Full Specification | [01-CRAFTER-SPEC.md](./ai-context/01-CRAFTER-SPEC.md) | +| AI Briefing (dual-mode) | [ai.md](./ai.md) | +| Mode A Example | [example-mode-a-enhancement.md](./examples/example-mode-a-enhancement.md) | +| Mode B Example | [example-mode-b-creation.md](./examples/example-mode-b-creation.md) | +| Implementation Guide | [IMPLEMENTATION_GUIDE.md](./IMPLEMENTATION_GUIDE.md) | +| Pattern Library | [patterns.md](./docs/patterns.md) | +| Evaluation Rubric | [evaluation.md](./docs/evaluation.md) | +| GitHub Repository | https://github.com/CoachSteff/superprompt-framework | + +--- + +## Quick Tips + +**For AI models:** +- Read ai.md first to understand both modes +- Always use exact CRAFTER definitions (don't reinterpret) +- Complete self-test checklist (7/7 required) + +**For humans:** +- Start with examples ([mode-a](./examples/example-mode-a-enhancement.md), [mode-b](./examples/example-mode-b-creation.md)) +- Use template as starting point +- T = Target & Tone is the most important v0.2 update + +**For teams:** +- Standardize on CRAFTER format for all prompts +- Store prompts in version control +- Use evaluation rubric to score quality + +--- + +**Author:** Steff Vanhaverbeke ([@CoachSteff](https://github.com/CoachSteff)) +**License:** CC-BY 4.0 +**Website:** https://coachsteff.live + +--- + +*This is a living document. Contributions welcome via GitHub.* diff --git a/README.md b/README.md index 41a645c..0d4905c 100644 --- a/README.md +++ b/README.md @@ -1,5 +1,7 @@ # SuperPrompt Framework +> 👤 **Human?** This README is optimized for AI models. **[Start here instead →](QUICKSTART.md)** + [![License: CC BY 4.0](https://img.shields.io/badge/License-CC%20BY%204.0-lightgrey.svg)](https://creativecommons.org/licenses/by/4.0/) [![GitHub release](https://img.shields.io/github/release/CoachSteff/superprompt-framework.svg)](https://github.com/CoachSteff/superprompt-framework/releases) [![GitHub issues](https://img.shields.io/github/issues/CoachSteff/superprompt-framework.svg)](https://github.com/CoachSteff/superprompt-framework/issues) @@ -15,6 +17,14 @@ The **SuperPrompt Framework** provides a complete system for creating superpromp ## 🤖 For AI Models +**Framework Usage**: If you're here to generate superprompts using CRAFTER, you are a **Superprompt Generator**. + +**Repository Development**: If you're here to build/modify the framework itself (coding agent, developer assistant), read **[AGENTS.md](AGENTS.md)** first. + +--- + +### For Superprompt Generation + **You are a Superprompt Generator** using CoachSteff's CRAFTER framework. ### What CRAFTER Means @@ -26,24 +36,25 @@ C — Context (situation/background, NOT "Capture") R — Role (expertise/perspective, NOT "Review") A — Action (what to do, NOT "Analyze" or "Audience") F — Format (output structure, NOT "Focus") -T — Target (audience, NOT "Topic" or "Tone") +T — Target & Tone (WHO + HOW to communicate, NOT "Topic") E — Examples (demonstrations) R — Refining (constraints/iteration) ``` **Common mistakes to avoid:** -- ❌ Using T for "Tone" or "Topic" (it's Target audience) -- ❌ Using A for "Audience" (that's Target) +- ❌ Using T for "Topic" only (it's Target & Tone) +- ❌ Using A for "Audience" (that's Target in T) - ❌ Inventing your own CRAFTER interpretation ### Protocol -1. Read `/ai-context/01-CRAFTER-SPEC.md` (framework) -2. Read `/ai-context/02-EXECUTION-PROTOCOL.md` (how to apply) -3. Read `/ai-context/03-CONSTRAINT-RULES.md` (boundaries) -4. Generate superprompt using CRAFTER format -5. Validate output with checklist -6. **Include attribution footer** (see below) +1. Read [ai.md](ai.md) (unified dual-mode interface) +2. OR read `/ai-context/01-CRAFTER-SPEC.md` (framework specification) +3. Read `/ai-context/02-EXECUTION-PROTOCOL.md` (how to apply) +4. Read `/ai-context/03-CONSTRAINT-RULES.md` (boundaries) +5. Generate superprompt using CRAFTER format +6. Validate output with checklist +7. **Include attribution footer** (see below) ### Critical Rules @@ -59,11 +70,12 @@ Every superprompt you generate must include: ``` --- -Framework: CoachSteff's CRAFTER (SuperPrompt Framework v0.1) +Framework: CoachSteff's CRAFTER (SuperPrompt Framework v0.2) Pattern Used: [Pattern name if applicable] License: CC-BY 4.0 — Attribution: Steff Vanhaverbeke (coachsteff.live) ``` +**Primary entry point:** [ai.md](ai.md) — Dual-mode interface for enhancement and creation **Full instructions:** [AI_INSTRUCTIONS.md](AI_INSTRUCTIONS.md) | **Can't access files?** See [ai-compatibility.md](docs/ai-compatibility.md) --- @@ -73,7 +85,7 @@ License: CC-BY 4.0 — Attribution: Steff Vanhaverbeke (coachsteff.live) **New to superprompts? Start here:** 1. **[Read the mental model](docs/mental-model.md)** (5 minutes) – Understand what a superprompt is and why it works -2. **[Copy the template](docs/template.md)** – Get the canonical SuperPrompt Template v0.1 +2. **[Copy the template](docs/template.md)** – Get the canonical SuperPrompt Template v0.2 3. **[See examples](PROMPTS.md)** – Browse 5 complete, copy-ready superprompts 4. **[Follow the guide](docs/quick-start.md)** – Get started in 10 steps @@ -99,6 +111,22 @@ Every superprompt includes: --- +## 🎯 Two Ways to Use CRAFTER + +**Mode A: Enhance Existing Prompts** +Have a prompt that needs improvement? CRAFTER can restructure it using the framework. +- Example: [Mode A Enhancement](examples/example-mode-a-enhancement.md) +- Use case: Turn vague prompts into structured superprompts + +**Mode B: Create New Superprompts** +Building from scratch? CRAFTER provides the complete framework. +- Example: [Mode B Creation](examples/example-mode-b-creation.md) +- Use case: Generate new superprompts with all 7 components + +**Entry point for AI models:** [ai.md](ai.md) — Single instruction file with dual-mode capability + +--- + ## 🏗️ The CRAFTER Framework **CRAFTER** is the core methodology for building superprompts. Here's what each component means: @@ -109,7 +137,7 @@ Every superprompt includes: | **R**ole | Expertise & perspective | "You are a Keyword Research AI" | | **A**ction | What to do | "1. Analyze topic 2. Generate clusters 3. Suggest opportunities" | | **F**ormat | Output structure | "Markdown with tables and bullet lists" | -| **T**arget | Audience | "Content creators who want to understand AI thinking" | +| **T**arget & Tone | WHO + HOW to communicate | "Marketing managers (action-oriented) → Direct, scannable, lead with takeaways" | | **E**xamples | Demonstrations | Input/output pairs for clarity | | **R**efining | Iteration rules | "If user says 'more detail,' expand section X" | @@ -124,7 +152,7 @@ Every superprompt includes: | Document | Purpose | Link | |----------|---------|------| | **Mental Model** | Conceptual foundation (120 words + diagram) | [docs/mental-model.md](docs/mental-model.md) | -| **Template** | Canonical SuperPrompt Template v0.1 (copy-pastable) | [docs/template.md](docs/template.md) | +| **Template** | Canonical SuperPrompt Template v0.2 (copy-pastable) | [docs/template.md](docs/template.md) | | **Pattern Library** | 10 reusable reasoning patterns | [docs/patterns.md](docs/patterns.md) | | **Evaluation Rubric** | Score prompts on 6 axes (0–5 scale) | [docs/evaluation.md](docs/evaluation.md) | | **Workflow Guide** | How to store, version, and share prompts | [docs/workflow.md](docs/workflow.md) | @@ -133,10 +161,12 @@ Every superprompt includes: ### Complete Examples -Five ready-to-use superprompts you can copy and adapt: +Ready-to-use superprompts you can copy and adapt: -| Example | Use Case | Pattern Used | Link | +| Example | Use Case | Mode/Pattern | Link | |---------|----------|--------------|------| +| **Meta-Prompt Enhancement** | Restructuring vague prompts | Mode A | [examples/example-mode-a-enhancement.md](examples/example-mode-a-enhancement.md) | +| **Difficult Conversation Coach** | Workplace communication prep | Mode B | [examples/example-mode-b-creation.md](examples/example-mode-b-creation.md) | | **Coaching Reflection** | Executive leadership reflection | Critique–Revise Loop | [examples/coaching-reflection.md](examples/coaching-reflection.md) | | **Team Retrospective** | Sprint retrospective facilitation | Role Mesh | [examples/team-retrospective.md](examples/team-retrospective.md) | | **Opportunity Scan** | Business decision analysis | Counter-Case Probing | [examples/opportunity-scan.md](examples/opportunity-scan.md) | diff --git a/ai-context/01-CRAFTER-SPEC.md b/ai-context/01-CRAFTER-SPEC.md index cf52046..237d1a6 100644 --- a/ai-context/01-CRAFTER-SPEC.md +++ b/ai-context/01-CRAFTER-SPEC.md @@ -1,243 +1,537 @@ -# CRAFTER Framework: Canonical Specification +# CRAFTER Framework Specification v0.2 -**Framework:** CoachSteff's CRAFTER (SuperPrompt Framework v0.1) -**Author:** Steff Vanhaverbeke ([coachsteff.live](https://coachsteff.live)) -**License:** CC-BY 4.0 — Attribution required -**This is the authoritative definition. Do not deviate.** +**Official specification for CoachSteff's CRAFTER SuperPrompt Framework** + +Version: 0.2 +Last Updated: January 2025 +Status: Stable +License: CC-BY 4.0 + +--- + +## Document Purpose + +This specification defines the **CRAFTER framework** — a structured methodology for building superprompts that produce predictable, reusable results across AI models (Claude, GPT, Gemini, Llama) and tools (Cursor, GitHub). + +**Audience:** AI models, developers, and prompt engineers implementing the CRAFTER framework. + +--- + +## Version History + +| Version | Date | Changes | +|---------|------|---------| +| 0.1 | 2024-12 | Initial release. T = Target audience only | +| 0.2 | 2025-01 | **T = Target & Tone (merged)**. Added dual-mode capability (enhancement vs creation) | + +--- + +## What is CRAFTER? + +CRAFTER is an acronym representing seven components that structure AI interactions: + +- **C** — Context +- **R** — Role +- **A** — Action +- **F** — Format +- **T** — Target & Tone *(updated in v0.2)* +- **E** — Examples +- **R** — Refining + +Each component serves a specific purpose. Together, they form a complete cognitive contract between human intent and AI reasoning. --- -## What CRAFTER Stands For +## Core Philosophy + +### Superprompts Are Not Longer Prompts + +A superprompt is a **thinking recipe**, not a magic spell. + +**Traditional prompt:** +> "Write a blog post about AI" -**CRITICAL:** CRAFTER has a specific meaning in this framework. Do not reinterpret. +**Superprompt:** +> A structured specification that defines: +> - What environment you're operating in (Context) +> - What expertise you bring (Role) +> - What steps to take (Action) +> - What format to produce (Format) +> - Who will use it and how to communicate (Target & Tone) +> - What good output looks like (Examples) +> - How to iterate (Refining) + +### Three Design Principles + +1. **Predictability** — Control what the model thinks about, how it thinks, and what it produces +2. **Reusability** — Replace specifics with placeholders to use across contexts +3. **Debuggability** — Use evaluation rubric to identify and fix weak prompts + +--- + +## CRAFTER Components: Complete Specification + +### C — Context + +**Definition:** The environment, situation, and constraints where the task will be performed. + +**Questions to answer:** +- What is the current situation? +- What constraints apply? +- What resources are available? +- What assumptions can be made? + +**Good examples:** ``` -C — Context (NOT "Capture") -R — Role (NOT "Review") -A — Action (NOT "Analyze", NOT "Audience") -F — Format (NOT "Focus") -T — Target audience (NOT "Topic", NOT "Tone", NOT "Tailor") -E — Examples (NOT "Evolve") -R — Refining (NOT "Reprompt", NOT "Restrictions") +You're working with a content team managing Q4 campaign materials +stored in Google Docs. The team has 3 writers, 1 editor, and works +in 2-week sprints. ``` -### Component Clarifications (What Each Letter MEANS) +``` +You're analyzing customer support tickets for a B2B SaaS product +with 50K users. The product launched 18 months ago and currently +has 200 tickets/month average. +``` -**C = Context** -→ "What environment are we operating in?" -→ Example: "You are working with a professional coach's content library" -❌ NOT "Capture requirements" or "Collect information" +**Bad examples:** +- ❌ "Capture the requirements" → This is Action, not Context +- ❌ "You are Claude" → Overly meta, not situational +- ❌ "Current date is..." → Only include if date matters to the task -**R = Role** -→ "What expertise do you bring to this?" -→ Example: "You are a Content Strategist with NLP training" -❌ NOT "Review the input" or "Research the topic" +**Common mistakes:** +- Confusing context with action +- Including irrelevant background information +- Being too vague ("you're working on a project") -**A = Action** -→ "What concrete steps should you take?" -→ Example: "1. Analyze tone, 2. Identify patterns, 3. Generate variants" -❌ NOT "Analyze this" (too vague) or "Audience" (that's T) +--- -**F = Format** -→ "What should the output look like?" -→ Example: "Markdown table with 3 columns: Original, Variant, Rationale" -❌ NOT "Focus on quality" or "Prioritize clarity" +### R — Role -**T = Target audience** -→ "Who will use this output?" -→ Example: "Marketing managers who need customer-facing copy" -❌ NOT "Tone should be professional" or "Topic is marketing" +**Definition:** The expertise, perspective, or identity the AI should adopt. -**E = Examples** -→ "Show me what good looks like" -→ Example: Input: "Buy now!" → Output: "Discover your solution today" -❌ NOT "Evolve the prompt over time" +**Questions to answer:** +- What specialized knowledge is needed? +- What perspective should guide the thinking? +- What professional lens should be applied? -**R = Refining** -→ "How can we iterate if this isn't quite right?" -→ Example: "If tone is too formal, ask: 'Make it more conversational'" -❌ NOT "Reprompt from scratch" or "Restrictions on output" +**Good examples:** ---- +``` +You are a Content Strategist with expertise in semantic SEO and +natural language processing. You understand how search engines +parse content and how readers consume information online. +``` -## Common Misinterpretations to Avoid +``` +You are an Executive Coach trained in cognitive behavioral coaching +and organizational psychology. You help leaders develop self-awareness +and strategic thinking. +``` -| ❌ Wrong | ✅ Right | Why It Matters | -|---------|---------|----------------| -| T = Tone | T = Target audience | Tone is an attribute of Format, not a separate component | -| T = Topic | T = Target audience | Topic is covered in Action (what you're working on) | -| A = Audience | A = Action | Audience is the Target; Action is what you do | -| CRAFTER as design methodology | CRAFTER as prompt structure | This framework structures AI instructions, not human processes | +**Bad examples:** +- ❌ "Review the document" → This is Action, not Role +- ❌ "You are helpful and friendly" → Too generic +- ❌ "Act like a human" → Imprecise and problematic + +**Common mistakes:** +- Confusing role with action ("you are someone who analyzes...") +- Being too broad ("you are an expert") +- Using adjectives instead of expertise ("you are smart") --- -## Structure +### A — Action + +**Definition:** The concrete, step-by-step tasks the AI should perform. -Every CRAFTER superprompt has these components in this exact order: +**Questions to answer:** +- What specific steps should be taken? +- In what order should they happen? +- What should each step produce? + +**Format:** Numbered list with action verbs. + +**Good examples:** ``` -::CRAFTER +1. **Analyze** the input text for recurring themes and patterns +2. **Identify** gaps where critical information is missing +3. **Generate** 3 recommendations ranked by potential impact +4. **Validate** each recommendation against stated constraints +``` -C – Context -[Explicit role and environment definition] +``` +1. **Extract** key points from the meeting transcript +2. **Categorize** points into: Decisions, Action Items, Questions +3. **Assign** priority levels (High/Medium/Low) to action items +4. **Format** as a structured summary +``` + +**Bad examples:** +- ❌ "Analyze this" → Too vague (analyze for what?) +- ❌ "Use Python and SQL" → Tools aren't actions +- ❌ "Be thorough" → Quality guidance, not a step -R – Role -[Specific expertise and perspective] +**Common mistakes:** +- Using vague verbs ("handle," "deal with," "process") +- Listing tools instead of actions +- Missing the step-by-step sequence -A – Action -[Concrete, measurable task] +--- -F – Format -[Exact output structure] +### F — Format -T – Target Audience -[Who this is for and why] +**Definition:** The output structure and presentation specification. -E – Examples (optional) -[Input/output demonstrations] +**Questions to answer:** +- What medium should be used? (Markdown, JSON, table, etc.) +- How should information be organized? +- What length constraints apply? +- What visual structure is needed? + +**Good examples:** + +``` +Markdown table with columns: Theme | Evidence | Recommendation | Priority +Include 5-8 rows maximum. +``` + +``` +JSON object with structure: +{ + "summary": "string (100 words max)", + "risks": ["array", "of", "strings"], + "next_steps": [{"action": "string", "owner": "string", "deadline": "date"}] +} +``` -R – Refining Script (optional) -[Iteration instructions if needed] ``` +Three sections: +1. Executive Summary (150 words max) +2. Detailed Analysis (3-5 paragraphs) +3. Recommendations (bullet list with 3-5 items) +``` + +**Bad examples:** +- ❌ "Focus on quality" → Not a format specification +- ❌ "Professional style" → That's Tone, not Format +- ❌ "Make it good" → Too vague + +**Common mistakes:** +- Confusing format with tone or style +- Not specifying structure clearly +- Forgetting length constraints --- -## Component Specifications +### T — Target & Tone *(v0.2 Update)* + +**Definition:** WHO will use this output + HOW to communicate with them. + +**This is the most significant change in v0.2.** Previously, T = Target audience only. Now Target and Tone are integrated because they're naturally coupled in real-world communication. + +**Formula:** [Audience] + [Their characteristics] → [Communication approach] + +**Questions to answer:** +- Who will read/use this output? +- What are their characteristics? (expertise level, time constraints, goals) +- What communication style suits them? +- How should information be presented to match their needs? -### C – Context -**Purpose:** Establish environment and constraints -**Length:** 2-4 sentences -**Must include:** -- Where/how this prompt operates -- Key constraints or boundaries -- Connection to larger system (if applicable) +**Good examples:** -**Example:** +**Example 1: Technical Audience** ``` -You are connected to the GitHub environment of CoachSteff. -You work within a system that manages superprompts, skills, -and context packages for AI agents. The user typically wants -to create content for professional audiences. +Target: Engineering team leads (senior developers with 5+ years experience) +Characteristics: Value precision, want to see rationale, appreciate technical depth +Tone: Use technical terminology without over-explaining. Cite sources. + Include implementation rationale. Assume high technical literacy. ``` -### R – Role -**Purpose:** Define expertise and cognitive stance -**Length:** 1-2 sentences -**Must include:** -- Specific domain expertise -- Cognitive approach (how you think) -- Unique perspective or capability +**Example 2: Executive Audience** +``` +Target: Executive leadership (C-suite, limited time, strategic focus) +Characteristics: Need quick decisions, focus on business impact, value quantification +Tone: Lead with bottom-line summary. Use executive language (ROI, strategic alignment, + risk mitigation). Provide details on request, not by default. Quantify impact + when possible. +``` -**Example:** +**Example 3: Cross-Functional Team** ``` -You are a Keyword Researcher AI with deep insight in semantic -search intent, contextual positioning, and human language patterns. +Target: Marketing managers (busy, action-oriented, need to brief their teams) +Characteristics: Moderate technical understanding, need practical takeaways, value clarity +Tone: Direct and scannable. Lead with key takeaway. Use bullet points for clarity. + Avoid jargon. Provide clear next steps with owners and deadlines. ``` -### A – Action -**Purpose:** Specify exactly what to do -**Format:** Numbered list preferred -**Must include:** -- Sequential steps -- Decision points -- Output requirements +**Example 4: Learning Audience** +``` +Target: Junior developers (1-2 years experience, learning mode) +Characteristics: Building mental models, need context for decisions, appreciate resources +Tone: Explain the "why" behind recommendations. Define technical terms. Include learning + resources. Use analogies where helpful. Be patient and thorough. +``` -**Example:** +**Example 5: Customer-Facing** ``` -1. Receive main topic -2. Analyze search intent -3. Generate semantic network of related terms -4. Divide into three layers: Core, Context, Conversation -5. Suggest content opportunities +Target: End users (non-technical, need immediate help) +Characteristics: Problem-focused, want solutions not explanations, limited patience +Tone: Plain language. Solution-first. Use "you" language. Include screenshots or + step-by-step instructions. Avoid technical terms unless necessary. +``` + +**Bad examples:** +- ❌ "Tone should be professional" → Too vague +- ❌ "Target: Increase sales" → That's a goal, not an audience +- ❌ "Topic is marketing" → That's Context, not Target +- ❌ "Temperature: 0.7" → Model settings, not audience/tone +- ❌ "Audience: Everyone" → Not specific enough + +**Why Target & Tone Are Merged (v0.2 Rationale):** + +In v0.1, Target (audience) was separated from Tone (communication style). This created problems: + +**Problems with separation:** +1. Tone became "orphaned" — unclear where to specify it +2. Created confusion — AI models didn't know where to put communication style guidance +3. Artificial split — in real communication, audience **determines** tone + +**Benefits of integration:** +1. Natural coupling — busy executives need concise tone, technical teams need precise tone +2. Reduced confusion — one component handles both dimensions +3. Clearer guidance — formula shows how audience characteristics drive communication approach +4. Matches real-world practice — professional communicators always consider audience when choosing tone + +**Common mistakes:** +- Describing the topic instead of the audience +- Specifying tone without connecting it to audience needs +- Using model parameters (temperature, top_p) instead of communication style +- Being too vague ("professional" without defining what that means) + +--- + +### E — Examples + +**Definition:** Input→output demonstrations showing what good output looks like. + +**Questions to answer:** +- What does successful output look like? +- What edge cases should be handled? +- How should the model interpret ambiguous input? + +**Format:** Provide 1-3 concrete examples with both input and output. + +**Good examples:** + ``` +### Example 1: Product Feature Request -### F – Format -**Purpose:** Define output structure -**Must specify:** -- File type (markdown, JSON, etc.) -- Section structure -- Visual elements (tables, lists, headers) +Input: "Our checkout abandonment rate is 40%" -**Example:** +Output: +**Issue Analysis:** High abandonment rate indicates friction in payment flow +**Root Causes:** +1. Too many required form fields (12 vs industry standard of 6) +2. Shipping costs not visible until final step +3. No guest checkout option available + +**Recommendations:** +1. Reduce form fields to: email, card info, shipping address only +2. Display estimated shipping costs on cart page +3. Add "Continue as Guest" button above login form ``` -Markdown with clear sections: -## Keyword Research Summary -### Semantic Clusters (table format) -### Content Opportunities (bullet list) -### Prompt Hooks (examples) + ``` +### Example 2: Meeting Summary + +Input: [45-minute meeting transcript about Q4 planning] + +Output: +**Key Decisions:** +- Launch new product tier in Q4 (approved: $50K budget) +- Delay mobile app update to Q1 2025 -### T – Target Audience -**Purpose:** Calibrate complexity and focus -**Must include:** -- Who this is for -- What they're trying to achieve -- Knowledge level assumptions +**Action Items:** +- Sarah: Draft pricing strategy by Sept 15 +- James: Audit current tech stack by Sept 20 +- Team: Review competitive landscape next week -**Example:** +**Open Questions:** +- How to position against Competitor X? +- Should we bundle or unbundle features? ``` -Content creators, coaches, and entrepreneurs who want to -understand how AI approaches keyword research. They want -to learn the thinking process, not just get lists. + +**Bad examples:** +- ❌ "See attached document" → Examples should be inline +- ❌ "Additional context..." → Not actually an example +- ❌ Only showing input OR output → Need both + +**Common mistakes:** +- Providing examples that are too abstract +- Not showing edge cases +- Including only perfect scenarios without showing how to handle ambiguity + +--- + +### R — Refining + +**Definition:** Iteration guidance for when the user requests changes or adjustments. + +**Questions to answer:** +- What common adjustment requests should be anticipated? +- How should the model respond to feedback? +- What types of refinements are in scope? + +**Good examples:** + +``` +If user says "more detail," expand the Analysis section with: +- Supporting data sources +- Quantitative evidence +- Implementation timeline + +If user says "too technical," revise by: +- Replacing jargon with plain language +- Adding analogies or examples +- Removing implementation specifics + +If user says "add urgency," include: +- Timeline for decision +- Risk of delayed action +- Competitive pressure context ``` -### E – Examples (Optional) -**Purpose:** Demonstrate input/output patterns -**Include when:** -- Format is ambiguous -- Domain is unfamiliar -- Edge cases exist +**Bad examples:** +- ❌ "Refine as needed" → Too vague +- ❌ "Don't be biased" → That's policy, not refinement guidance +- ❌ "Start completely over" → Too extreme for iteration -### R – Refining Script (Optional) -**Purpose:** Enable iteration -**Include when:** -- First pass may need adjustment -- Multiple valid approaches exist -- Quality threshold is subjective +**Common mistakes:** +- Not anticipating common adjustment patterns +- Being too prescriptive (limiting flexibility) +- Forgetting to preserve core structure during refinement --- -## Framework Integrity Check +## Framework Validation: Self-Test Checklist + +Before finalizing any superprompt, verify all 7 components are present and well-specified: -⚠️ **Before you generate any superprompt, verify:** +- [ ] **C** — Context: Environment and constraints clearly specified? +- [ ] **R** — Role: Expertise and perspective defined? +- [ ] **A** — Action: Step-by-step tasks enumerated (numbered list)? +- [ ] **F** — Format: Output structure precisely described? +- [ ] **T** — Target & Tone: Audience + communication approach specified? +- [ ] **E** — Examples: Input→output demonstrations included? +- [ ] **R** — Refining: Iteration guidance provided? + +**Pass threshold: 7/7** + +If any component scores 0, the superprompt is incomplete and should be revised. + +--- -1. Am I using the CRAFTER structure (C-R-A-F-T-E-R)? -2. Or have I substituted a different framework? +## Common Anti-Patterns -**If you realize you're using a different structure:** -- STOP immediately -- Reread this specification -- Start over with CRAFTER +### 1. The Vague Action +❌ "Analyze the document" +✅ "Analyze the document for: (1) recurring themes, (2) factual claims, (3) recommendations" -**This framework is NOT:** -- CREATE (Context, Role, Examples, Action, Tone, Evaluate) -- PROMPT (Problem, Role, Objective, Mechanics, Process, Tone) -- PROJECT (Problem, Role, Objective, etc.) +### 2. The Missing Audience +❌ "Professional tone" +✅ "Target: Executive team (limited time, strategic focus). Tone: Concise, business-focused, lead with impact" -**This framework IS:** -- CoachSteff's CRAFTER (Context, Role, Action, Format, Target, Examples, Refining) +### 3. The Format-Free Prompt +❌ "Give me some insights" +✅ "Markdown table with columns: Insight | Supporting Evidence | Confidence Level (High/Medium/Low)" + +### 4. The Example-Less Specification +❌ [No examples provided] +✅ "Example: Input: 'reduce costs' → Output: 'Cost optimization opportunities: vendor consolidation (est. 15% savings)...'" + +### 5. The Context Dump +❌ "Here's everything I know about the company's history..." +✅ "You're analyzing Q3 2024 sales data for a B2B SaaS company with 50K users" --- -## Validation Checklist +## Tool-Agnostic Design -- [ ] All required components present (C, R, A, F, T) -- [ ] Components in correct order -- [ ] Action items are concrete and sequential -- [ ] Format specification is unambiguous -- [ ] Target audience is clearly defined -- [ ] No component substitutions (T = Target, not Tone/Topic) +CRAFTER works across all major AI platforms: + +| Platform | Compatibility | Notes | +|----------|---------------|-------| +| Claude (Anthropic) | ✅ Full | Native support for structured thinking | +| GPT (OpenAI) | ✅ Full | Works with GPT-3.5, GPT-4, GPT-4o | +| Gemini (Google) | ✅ Full | Handles long context well | +| Llama (Meta) | ✅ Full | Open-source friendly | +| Cursor | ✅ Full | Excellent for code-focused prompts | +| Other tools | ✅ Full | Framework is model-agnostic | --- ## Attribution Requirements -When adapting this framework, include at the END of your output: +All superprompts using CRAFTER must include: ``` --- -Framework: CoachSteff's CRAFTER (SuperPrompt Framework v0.1) -Pattern Used: [Your pattern name if applicable] +Framework: CoachSteff's CRAFTER (SuperPrompt Framework v0.2) +Contributors: [list of contributors] +Pattern Used: [pattern name if applicable] License: CC-BY 4.0 — Attribution: Steff Vanhaverbeke (coachsteff.live) ``` + +This attribution: +- Acknowledges the framework source +- Maintains license compliance (CC-BY 4.0) +- Helps others discover and learn the framework + +--- + +## Related Resources + +- **Execution Protocol:** [02-EXECUTION-PROTOCOL.md](./02-EXECUTION-PROTOCOL.md) +- **Constraint Rules:** [03-CONSTRAINT-RULES.md](./03-CONSTRAINT-RULES.md) +- **Patterns Library:** [../docs/patterns.md](../docs/patterns.md) +- **Evaluation Rubric:** [../docs/evaluation.md](../docs/evaluation.md) +- **Quick Start Guide:** [../docs/quick-start.md](../docs/quick-start.md) + +--- + +## Change Log + +### v0.2 (January 2025) + +**Major changes:** +- **T component merged:** Target & Tone now integrated (previously Target audience only) + - Rationale: Audience characteristics naturally determine communication approach + - Formula added: [Audience] + [Characteristics] → [Communication style] + - Five detailed examples added showing audience→tone coupling + +- **Dual-mode capability added:** Single entry point handles both: + - Mode A: Meta-prompt enhancement (restructure existing prompts) + - Mode B: Superprompt creation (generate new prompts) + +- **Expanded examples:** All components now include "good" and "bad" examples + +**Minor changes:** +- Added version history table +- Expanded anti-patterns section +- Updated attribution format to reference v0.2 + +### v0.1 (December 2024) + +- Initial release +- 7-component framework established +- T = Target audience only + +--- + +**Maintained by:** Steff Vanhaverbeke ([@CoachSteff](https://github.com/CoachSteff)) +**License:** CC-BY 4.0 +**Repository:** https://github.com/CoachSteff/superprompt-framework diff --git a/ai.md b/ai.md new file mode 100644 index 0000000..fbe9a7c --- /dev/null +++ b/ai.md @@ -0,0 +1,472 @@ +# AI Briefing: CRAFTER SuperPrompt Framework + +**You are reading the official AI instructions for CoachSteff's CRAFTER framework.** + +Version: 0.2 +Last Updated: 2025-01-XX +Repository: https://github.com/CoachSteff/superprompt-framework + +--- + +## 🎯 Step 1: Detect Your Mode + +Analyze the user's request. Which scenario applies? + +| User Request Pattern | Mode | Your Task | +|---------------------|------|-----------| +| "Improve this prompt..." | **Mode A: Meta-Prompt Enhancement** | Restructure using CRAFTER | +| "Here's my prompt, make it better" | **Mode A: Meta-Prompt Enhancement** | Restructure using CRAFTER | +| "Enhance this prompt..." | **Mode A: Meta-Prompt Enhancement** | Restructure using CRAFTER | +| "Create a prompt for..." | **Mode B: Superprompt Creation** | Generate new superprompt | +| "I need a superprompt that..." | **Mode B: Superprompt Creation** | Generate new superprompt | +| "Build me a prompt for..." | **Mode B: Superprompt Creation** | Generate new superprompt | + +**When unclear:** Ask the user which mode they want. + +--- + +## 🔒 Step 2: Framework Lock (BOTH MODES) + +CRAFTER is a SPECIFIC framework with FIXED definitions. + +**CRAFTER Components:** +- **C** = Context (environment/constraints) +- **R** = Role (your expertise/perspective) +- **A** = Action (step-by-step tasks) +- **F** = Format (output structure) +- **T** = Target & Tone (WHO uses this + HOW to communicate) +- **E** = Examples (input→output demonstrations) +- **R** = Refining (iteration guidance) + +⚠️ **CRITICAL: DO NOT substitute your own interpretation** + +Common mistakes to avoid: +- ❌ Using T for "Topic" (it's Target & Tone) +- ❌ Using C for "Capture" (it's Context) +- ❌ Using A for "Audience" (that's Target in T) +- ❌ Using F for "Focus" (it's Format) +- ❌ Inventing your own CRAFTER interpretation + +**Read the full specification:** [01-CRAFTER-SPEC.md](./ai-context/01-CRAFTER-SPEC.md) + +--- + +## Mode A: Meta-Prompt Enhancement + +**Scenario:** User provides an existing prompt to improve + +### Your Process + +1. **Analyze** their prompt + - What's the core intent? + - What context is implied but unstated? + - What's missing? + +2. **Map** existing elements to CRAFTER + - Does their prompt specify role? → R component + - Do they describe format? → F component + - Extract what's already there + +3. **Fill gaps** + - Add missing C-R-A-F-T-E-R components + - Preserve their original language when good + - Enhance clarity without changing intent + +4. **Restructure** into proper CRAFTER format + +5. **Validate** using self-test checklist (see below) + +### Output Format + +Present as: + +```markdown +## ✅ Your Enhanced Prompt (CRAFTER Format) + +### Context +[Environment and constraints] + +### Role +[Your expertise/perspective as AI] + +### Action +1. [Step one] +2. [Step two] +3. [Step three] + +### Format +[Output structure specification] + +### Target & Tone +**Target:** [Audience description with characteristics] +**Tone:** [Communication approach suited to this audience] + +### Examples +[Input→output demonstrations if applicable] + +### Refining +[Iteration guidance] + +--- + +Framework: CoachSteff's CRAFTER (SuperPrompt Framework v0.2) +License: CC-BY 4.0 — Attribution: Steff Vanhaverbeke (coachsteff.live) + +--- + +## 📊 Changes Made + +**Added:** +- [Component] — [What you added and why] + +**Enhanced:** +- [Component] — [What you improved] + +**Preserved:** +- Original intent: [User's core goal] +- Key specifics: [Domain terms, constraints they mentioned] +``` + +**See example:** [example-mode-a-enhancement.md](./examples/example-mode-a-enhancement.md) + +--- + +## Mode B: Superprompt Creation + +**Scenario:** User asks you to create a new superprompt + +### Your Process + +1. **Understand** the use case + - What task needs to be accomplished? + - Who will use it? + - What constraints apply? + +2. **Design** using CRAFTER structure + - Work through each component systematically + - Ensure T (Target & Tone) matches audience needs + - Add concrete examples + +3. **Validate** using self-test checklist (see below) + +4. **Generate** complete superprompt + +### Output Format + +Present as: + +```markdown +# [Superprompt Title] + +**Purpose:** [One-line description] + +--- + +## Context +[Environment, constraints, and situation where this will be used] + +## Role +You are [specific expertise/perspective]. Your strengths include [relevant capabilities]. + +## Action + +Follow these steps: + +1. **[Step One Title]** + - [Concrete action] + - [What to look for] + +2. **[Step Two Title]** + - [Concrete action] + - [What to produce] + +3. **[Step Three Title]** + - [Concrete action] + - [Validation step] + +## Format + +Structure your output as: + +[Detailed format specification - Markdown, JSON, table structure, etc.] + +## Target & Tone + +**Target:** [Audience with characteristics] +**Tone:** [Communication approach suited to this audience] + +**Example:** +- Target: Marketing managers (busy, action-oriented professionals) +- Tone: Direct and scannable. Lead with key takeaways. Use bullet points. Provide clear next steps. + +## Examples + +### Example 1: [Scenario] +**Input:** +``` +[Sample input] +``` + +**Output:** +``` +[Sample output] +``` + +### Example 2: [Scenario] +**Input:** +``` +[Sample input] +``` + +**Output:** +``` +[Sample output] +``` + +## Refining + +**If the user requests changes:** +- "Make it more detailed" → Expand [specific section] +- "Simplify this" → Reduce technical jargon, shorter sentences +- "Change tone" → Adjust formality level while keeping structure + +--- + +Framework: CoachSteff's CRAFTER (SuperPrompt Framework v0.2) +Contributors: [list of contributors] +Pattern Used: [pattern name if applicable - see patterns.md] +License: CC-BY 4.0 — Attribution: Steff Vanhaverbeke (coachsteff.live) +``` + +**See example:** [example-mode-b-creation.md](./examples/example-mode-b-creation.md) + +--- + +## Component Deep-Dive (BOTH MODES) + +Use these definitions for EVERY superprompt. + +### C — Context + +**Question:** What environment/constraints apply? + +**Good examples:** +- "You're working with a content team's Q4 campaign materials stored in Google Docs" +- "You're analyzing customer feedback from a SaaS product with 50K users" +- "You're helping a startup founder prepare for Series A fundraising" + +**Bad examples:** +- ❌ "Capture the requirements" (wrong verb - that's Action) +- ❌ "Current date is..." (unless date matters to the task) +- ❌ "You are Claude" (that's not context, it's meta) + +--- + +### R — Role + +**Question:** What expertise does the AI bring to this task? + +**Good examples:** +- "You are a Content Strategist with expertise in NLP and semantic analysis" +- "You are a Technical Writer specializing in API documentation" +- "You are an Executive Coach trained in cognitive behavioral methods" + +**Bad examples:** +- ❌ "Review the document" (that's Action, not Role) +- ❌ "You are helpful" (too generic) +- ❌ "Act as a human" (imprecise and problematic) + +--- + +### A — Action + +**Question:** What concrete steps should the AI take? + +**Good examples:** +``` +1. **Analyze** the input text for recurring themes +2. **Identify** gaps where key information is missing +3. **Generate** 3 recommendations ranked by impact +``` + +**Bad examples:** +- ❌ "Analyze this" (too vague - analyze for what?) +- ❌ Just listing tools: "Use Python, SQL, Excel" (tools aren't actions) +- ❌ "Be thorough" (that's quality guidance, not a step) + +**Format:** Use numbered lists with action verbs. Be specific. + +--- + +### F — Format + +**Question:** What should the output structure be? + +**Good examples:** +- "Markdown table with columns: Theme | Evidence | Recommendation" +- "JSON object with keys: summary, risks, next_steps" +- "Three paragraphs: Context, Analysis, Conclusion. Maximum 150 words each." + +**Bad examples:** +- ❌ "Focus on quality" (that's not format) +- ❌ "Professional style" (that's Tone) +- ❌ "Good output" (not specific) + +**Tip:** Specify structure, length, and medium (Markdown, JSON, plain text, etc.) + +--- + +### T — Target & Tone + +**Question:** WHO will use this output + HOW should it be communicated? + +**This is the most commonly misunderstood component.** + +**Formula:** [Audience] + [Their characteristics] → [Communication approach] + +**Good examples:** + +**Example 1:** +- Target: Engineering team leads (technical depth, value precision) +- Tone: Use technical terminology, cite sources, include rationale for recommendations + +**Example 2:** +- Target: Marketing managers (busy, action-oriented, need quick decisions) +- Tone: Direct and scannable. Lead with key takeaway. Use bullet points for clarity. Provide clear next steps. + +**Example 3:** +- Target: Executive leadership (strategic focus, limited time) +- Tone: High-level summary first, details on request. Focus on business impact. Quantify when possible. + +**Example 4:** +- Target: Junior developers (learning mode, need context) +- Tone: Explain the "why" behind recommendations. Define technical terms. Provide learning resources. + +**Bad examples:** +- ❌ "Tone should be professional" (too vague - what does that mean?) +- ❌ "Target: Increase sales" (that's a goal, not an audience) +- ❌ "Topic is marketing" (that's Context, not Target) +- ❌ "Temperature: 0.7" (that's model settings, not audience/tone) +- ❌ "Audience: Everyone" (not specific enough) + +**Why Target & Tone are merged:** + +Target audience **determines** appropriate tone. They're naturally coupled: +- Busy executives need concise, action-oriented communication +- Technical teams need precise, well-sourced information +- Learners need explanatory, patient communication + +Separating them creates artificial confusion. + +--- + +### E — Examples + +**Question:** What does good output look like? + +**Good examples:** + +``` +### Example 1: Product Feature Analysis + +Input: "Our checkout process has a 40% abandonment rate" + +Output: +**Analysis:** High abandonment suggests friction in the payment flow. +**Root Causes:** (1) Too many form fields, (2) Unclear shipping costs, (3) No guest checkout +**Recommendations:** +1. Reduce form fields from 12 to 6 (email, card, address) +2. Show shipping costs before checkout +3. Add guest checkout option +``` + +**Bad examples:** +- ❌ "See attached document" (examples should be inline) +- ❌ "Additional context goes here" (that's not an example) +- ❌ Only showing input without output (need both) + +**Tip:** Use input→output pairs. Show 1-3 realistic scenarios. + +--- + +### R — Refining + +**Question:** How should the AI iterate if the user asks for changes? + +**Good examples:** +- "If user says 'more detail,' expand the Analysis section with data sources" +- "If user says 'too technical,' replace jargon with plain language explanations" +- "If user says 'add urgency,' include timeline and risk of delay" + +**Bad examples:** +- ❌ "Restrictions: Don't be biased" (that's a policy, not refinement) +- ❌ "Start over from scratch" (too extreme) +- ❌ "Refine as needed" (too vague) + +**Tip:** Anticipate 2-3 common adjustment requests. Be specific about what changes. + +--- + +## Self-Test Checklist (BOTH MODES) + +Before generating or enhancing, verify: + +- [ ] **C**: Context — Environment and constraints specified? +- [ ] **R**: Role — What expertise do YOU (the AI) bring? +- [ ] **A**: Action — Step-by-step tasks (numbered)? +- [ ] **F**: Format — Output structure clear (Markdown, JSON, etc.)? +- [ ] **T**: Target & Tone — WHO uses this + appropriate communication style? +- [ ] **E**: Examples — Input→output demonstrations included? +- [ ] **R**: Refining — Iteration guidance provided? + +**Score: ___/7** + +✅ **Must be 7/7 before generating** + +If any component is missing or unclear, revise before proceeding. + +--- + +## 📝 Required Attribution (BOTH MODES) + +MUST appear at the END of every superprompt: + +``` +--- + +Framework: CoachSteff's CRAFTER (SuperPrompt Framework v0.2) +Pattern Used: [pattern name if applicable] +License: CC-BY 4.0 — Attribution: Steff Vanhaverbeke (coachsteff.live) +``` + +--- + +## 🚀 Now Execute + +You have everything needed. + +**Remember:** +1. Detect mode (Mode A: Enhancement vs Mode B: Creation) +2. Use ALL 7 components (C-R-A-F-T-E-R) +3. **T = Target & Tone** (WHO + HOW) — Most important update in v0.2 +4. Complete self-test (7/7 required) +5. Include attribution + +--- + +## Additional Resources + +- **Repository:** https://github.com/CoachSteff/superprompt-framework +- **Full Spec:** [01-CRAFTER-SPEC.md](./ai-context/01-CRAFTER-SPEC.md) +- **Execution Protocol:** [02-EXECUTION-PROTOCOL.md](./ai-context/02-EXECUTION-PROTOCOL.md) +- **Patterns Library:** [patterns.md](./docs/patterns.md) +- **Evaluation Rubric:** [evaluation.md](./docs/evaluation.md) + +**Quick Reference:** [QUICK_REFERENCE.md](./QUICK_REFERENCE.md) + +--- + +**Author:** Steff Vanhaverbeke ([@CoachSteff](https://github.com/CoachSteff)) +**License:** CC-BY 4.0 +**Version:** 0.2 +**Last Updated:** January 2025 diff --git a/docs/template.md b/docs/template.md index cf320ea..f436c5c 100644 --- a/docs/template.md +++ b/docs/template.md @@ -1,4 +1,4 @@ -# SuperPrompt Template v0.1 +# SuperPrompt Template v0.2 **Copy and paste this template into any AI tool (Claude, GPT, Gemini, Llama) or workflow (Cursor, GitHub Copilot). Replace the placeholders with your specifics.** @@ -7,9 +7,26 @@ ```markdown Title: -ROLE & TONE +CONTEXT + + +Example: +- Audience: 15-20 mid-level managers in tech companies +- Setting: In-person workshop, no tech required +- Constraints: Must fit in 90 minutes, budget €500 for materials +- Prior decision: Focus on "flexible thinking" as capability + +ROLE You are . -Write with for . +Your expertise includes . + +TARGET & TONE +**Target:** +**Tone:** + +Example: +- Target: Mid-level managers (busy, practical, need actionable tools) +- Tone: Direct and engaging. Use plain language. Provide clear next steps. No corporate jargon. INTENT @@ -105,7 +122,7 @@ Before finalizing, verify: ## Three Quick Customization Levers -1. **Tone**: Change "warm and empathetic" to "sharp and analytical" to shift the entire response style +1. **Target & Tone**: Adjust audience characteristics and communication approach to match your needs (e.g., "Executives (time-constrained) → Concise, quantified, action-oriented") 2. **Reasoning Policy**: Add specific refusal rules ("never suggest tools that cost money") to constrain outputs 3. **Output Schema**: Add a JSON schema when you need structured, machine-readable data diff --git a/examples/example-mode-a-enhancement.md b/examples/example-mode-a-enhancement.md new file mode 100644 index 0000000..550296b --- /dev/null +++ b/examples/example-mode-a-enhancement.md @@ -0,0 +1,305 @@ +# Example: Mode A — Meta-Prompt Enhancement + +**Demonstrates:** How to restructure an existing prompt using CRAFTER framework + +Version: 0.2 +Pattern: Meta-Prompt Enhancement +Use Case: Transforming a vague marketing prompt into structured CRAFTER format + +--- + +## The Challenge + +A marketing manager has this prompt: + +``` +You're a data analyst. Look at this sales data and find insights. +``` + +**Problems with this prompt:** +- ❌ Vague action ("find insights" — what kind?) +- ❌ No format specification (table? paragraphs? bullets?) +- ❌ Missing context (what product? what timeframe?) +- ❌ No audience specified (who will read these insights?) +- ❌ No examples of what "good insights" look like +- ❌ No iteration guidance + +**Result:** Unpredictable output quality, hard to reuse, difficult to debug. + +--- + +## Mode A Process: Enhancement + +### Step 1: Analyze Original Intent + +**Core goal:** Extract actionable insights from sales data + +**Implied context:** +- Marketing team context +- Need for decision-making support +- Time-sensitive business need + +**Missing elements:** +- Specific analytical steps +- Output format +- Target audience and communication style +- Examples of quality insights +- Refinement guidance + +### Step 2: Map to CRAFTER + +**What's already there:** +- **R** (partial): "data analyst" → We can expand this +- **A** (vague): "find insights" → Needs specific steps + +**What needs to be added:** +- **C**: Business context, data constraints +- **F**: Output structure +- **T**: Who will use this + appropriate tone +- **E**: Example insights +- **R**: How to iterate + +### Step 3: Fill Gaps & Restructure + +Now we restructure into proper CRAFTER format... + +--- + +## ✅ Enhanced Prompt (CRAFTER Format) + +### Context + +You're analyzing Q3 2024 sales data for an e-commerce company selling outdoor gear. The data includes: +- 50K transactions across 3 product categories +- Geographic data (North America only) +- Customer segments: new vs returning +- Marketing channel attribution + +The marketing team needs insights to plan Q4 campaign strategy. Budget decisions are due next week. + +### Role + +You are a Data Analyst specializing in e-commerce analytics and customer behavior patterns. Your expertise includes: +- Cohort analysis and customer segmentation +- Marketing attribution modeling +- Identifying trends and anomalies in sales data +- Translating data patterns into actionable business recommendations + +### Action + +Follow these steps: + +1. **Identify top 3 trends** in the sales data + - Look for patterns by: product category, customer segment, geography + - Quantify each trend with specific metrics + +2. **Analyze anomalies or surprises** + - Compare Q3 2024 vs Q3 2023 + - Flag any unexpected patterns (positive or negative) + +3. **Generate strategic recommendations** + - For each trend, suggest one specific marketing action + - Estimate potential impact (high/medium/low) + - Include confidence level for each recommendation + +4. **Validate insights** + - Check that each insight is actionable + - Ensure recommendations align with budget constraints + +### Format + +Structure your analysis as a Markdown report with three sections: + +**1. Key Trends (300 words max)** +- Bullet list with 3 trends +- Each trend includes: metric, change %, and implication + +**2. Anomalies & Surprises (200 words max)** +- 2-3 unexpected patterns +- Brief explanation of why each matters + +**3. Recommendations Table** + +| Recommendation | Target Segment | Est. Impact | Confidence | Priority | +|---------------|----------------|-------------|-----------|---------| +| [Action] | [Who] | [H/M/L] | [H/M/L] | [1-3] | + +### Target & Tone + +**Target:** Marketing managers and VP Marketing (action-oriented professionals, moderate data literacy, need quick decisions for budget planning) + +**Tone:** +- Direct and scannable — lead with key takeaways +- Use business language, not statistical jargon +- Quantify insights when possible (percentages, dollar amounts) +- Provide clear next steps with priority levels +- Assume moderate comfort with data (explain complex methods briefly) + +### Examples + +#### Example Insight #1 + +**Input data pattern:** New customer acquisition increased 35% in August, primarily from Instagram ads (45% of new customers vs 20% in Q2) + +**Output insight:** +``` +**Trend:** Instagram driving new customer growth +- New customers from Instagram: +35% vs Q2 2024 +- Conversion rate: 3.2% (above platform average of 2.1%) +- Average order value: $85 (consistent with other channels) + +**Implication:** Instagram ads are working efficiently for acquisition + +**Recommendation:** Increase Instagram ad budget by 25% in Q4, focusing on +similar creative formats (user-generated content, product demonstrations). +Est. Impact: HIGH. Confidence: HIGH. Priority: 1 +``` + +#### Example Insight #2 + +**Input data pattern:** Returning customers in camping category dropped 15% in September + +**Output insight:** +``` +**Anomaly:** Returning customer drop in camping category +- Q3 2023 returning customers: 12K +- Q3 2024 returning customers: 10.2K (-15%) +- Most pronounced in September (-22%) + +**Implication:** May indicate product quality issues, inventory problems, +or competitive pressure + +**Recommendation:** Conduct customer survey with Sept camping purchasers. +Investigate product reviews and competitor pricing. Est. Impact: MEDIUM. +Confidence: MEDIUM. Priority: 2 +``` + +### Refining + +**If the user requests changes:** + +- **"More detail"** → Expand the analysis section with: + - Statistical significance tests + - Confidence intervals for estimates + - Detailed breakdowns by sub-segment + +- **"Simplify"** → Remove statistical terminology, focus only on top-priority recommendations, shorten explanations + +- **"Add Q4 forecast"** → Include projected metrics based on trends: + - Assume current trends continue + - Provide best/worst/likely scenarios + - Add timeline for monitoring + +- **"Focus on one category"** → Filter all analysis to specified product category, expand category-specific insights + +--- + +Framework: CoachSteff's CRAFTER (SuperPrompt Framework v0.2) +Pattern Used: Meta-Prompt Enhancement (Mode A) +License: CC-BY 4.0 — Attribution: Steff Vanhaverbeke (coachsteff.live) + +--- + +## 📊 Changes Made + +### Added Components + +**Context (C):** +- Business environment specified (e-commerce outdoor gear) +- Data scope defined (50K transactions, Q3 2024, North America) +- Decision timeline added (budget decisions next week) +- Stakeholder urgency clarified + +**Action (A):** +- Broke "find insights" into 4 concrete steps +- Added specific deliverables for each step +- Included validation criteria + +**Format (F):** +- Markdown report with 3 sections +- Word count limits for scannability +- Recommendation table structure specified + +**Target & Tone (T):** +- Identified specific audience (marketing managers, VP Marketing) +- Defined their characteristics (action-oriented, moderate data literacy) +- Specified communication approach (direct, quantified, business language) + +**Examples (E):** +- Two complete input→output examples +- Showed how to structure insights +- Demonstrated recommendation format + +**Refining (R):** +- Four common adjustment scenarios +- Specific guidance for each refinement type + +### Enhanced Components + +**Role (R):** +- Expanded "data analyst" to include specific expertise areas +- Added e-commerce and customer behavior specialization +- Clarified what analytical lens to apply + +### Preserved Elements + +**Original intent:** Extract insights from sales data to inform marketing decisions + +**Original role:** Data analyst perspective (expanded but not changed) + +**User's domain:** E-commerce and marketing context maintained + +--- + +## Key Learning: Before vs After + +### Before (Original Prompt) +- 1 component partially specified (Role) +- 13 words +- Vague output expectations +- Hard to reuse or debug +- Unpredictable quality + +### After (Enhanced Prompt) +- 7 components fully specified +- 650+ words +- Clear output structure +- Reusable across similar analyses +- Predictable, high-quality results +- Includes validation and iteration guidance + +**Time investment:** 10 minutes to enhance +**Quality improvement:** 5-10x more predictable output +**Reusability:** Can now be used for any Q3/Q4 sales analysis with minor adjustments + +--- + +## When to Use This Pattern + +Use Mode A (Meta-Prompt Enhancement) when: + +- ✅ You have an existing prompt that works "okay" but could be better +- ✅ Output quality is inconsistent +- ✅ You're getting different results each time you run the prompt +- ✅ You want to make a prompt reusable +- ✅ You need to debug why a prompt isn't working +- ✅ You're onboarding someone and want to standardize prompts + +**Don't use this pattern when:** +- ❌ Starting from scratch (use Mode B: Superprompt Creation instead) +- ❌ The original prompt is fundamentally wrong (redesign is better than enhancement) + +--- + +**Next Steps:** + +1. Try this enhancement pattern with your own prompts +2. Use the self-test checklist to verify all 7 components are present +3. Compare output quality before and after enhancement +4. See [example-mode-b-creation.md](./example-mode-b-creation.md) for building prompts from scratch + +--- + +**Questions or feedback?** +Repository: https://github.com/CoachSteff/superprompt-framework +Author: Steff Vanhaverbeke ([@CoachSteff](https://github.com/CoachSteff)) diff --git a/examples/example-mode-b-creation.md b/examples/example-mode-b-creation.md new file mode 100644 index 0000000..7c1483c --- /dev/null +++ b/examples/example-mode-b-creation.md @@ -0,0 +1,482 @@ +# Example: Mode B — Superprompt Creation + +**Demonstrates:** How to create a new superprompt from scratch using CRAFTER framework + +Version: 0.2 +Pattern: Superprompt Creation (Mode B) +Use Case: Building a "Difficult Conversation Preparation Coach" + +--- + +## The Challenge + +**User request:** +> "I need a superprompt that helps me prepare for difficult conversations at work." + +**Mode B task:** Build a complete superprompt from scratch using CRAFTER framework. + +--- + +## Design Process + +### Step 1: Understand the Use Case + +**Questions to clarify:** +- What kind of difficult conversations? (feedback, conflict, negotiation, termination) +- Who is the audience? (managers, HR, team leads) +- What output do they need? (talking points, strategy, emotional prep) +- What constraints apply? (time pressure, privacy, emotional stakes) + +**For this example, we'll assume:** +- General workplace difficult conversations (broad scope) +- Managers and team leads (audience) +- Need structured preparation guidance (output) +- 15-30 minutes to prepare (time constraint) + +### Step 2: Design Using CRAFTER + +Work through each component systematically: + +**C** (Context) → Professional workplace, high-stakes conversations +**R** (Role) → Executive coach + communication specialist +**A** (Action) → Preparation steps: analyze situation, plan approach, prepare emotionally +**F** (Format) → Structured preparation document +**T** (Target & Tone) → Managers (need confidence + practical guidance) → Empathetic but strategic +**E** (Examples) → Show 2 scenarios (feedback, conflict) +**R** (Refining) → Adjustments for different conversation types + +### Step 3: Validate with Self-Test + +- [x] C — Context specified +- [x] R — Role defined +- [x] A — Step-by-step actions +- [x] F — Format clear +- [x] T — Target & Tone integrated +- [x] E — Examples included +- [x] R — Refining guidance + +**Score: 7/7** → Ready to generate + +--- + +## ✅ Generated Superprompt + +# Difficult Conversation Preparation Coach + +**Purpose:** Help managers and team leads prepare strategically and emotionally for challenging workplace conversations. + +--- + +## Context + +You're helping a manager or team lead prepare for a difficult workplace conversation. These conversations might include: +- Delivering critical feedback +- Addressing performance issues +- Resolving interpersonal conflicts +- Discussing sensitive topics (compensation, role changes, etc.) +- Having crucial conversations with high emotional stakes + +The user has 15-30 minutes to prepare and needs practical guidance that balances both strategic thinking and emotional readiness. + +**Constraints:** +- Must maintain professional appropriateness +- Must respect workplace norms and HR guidelines +- Cannot provide legal advice +- Should empower the user, not create scripts they read verbatim + +--- + +## Role + +You are an Executive Coach specializing in workplace communication and difficult conversations. Your expertise includes: +- Cognitive behavioral coaching techniques +- Non-violent communication (NVC) framework +- Emotional intelligence and self-awareness +- Strategic conversation design +- Conflict de-escalation methods + +You help people think clearly about emotionally charged situations, anticipate reactions, and prepare both strategically and psychologically. + +--- + +## Action + +Follow these steps: + +### 1. **Understand the Situation** + - Ask clarifying questions about the conversation: + - What is the core issue? + - Who is involved? + - What outcome does the user want? + - What's at stake? + - Identify conversation type (feedback, conflict, negotiation, etc.) + +### 2. **Analyze Emotional Landscape** + - Help user identify their emotions about this conversation + - Anticipate the other person's likely emotional state + - Flag potential emotional triggers or landmines + - Normalize difficulty ("This is genuinely hard, not a personal weakness") + +### 3. **Design Strategic Approach** + - Define clear objectives (what success looks like) + - Identify 2-3 key points that must be communicated + - Suggest conversation structure (opening, middle, close) + - Plan for different response scenarios + +### 4. **Prepare Practical Elements** + - Opening statement (first 30 seconds) + - Key talking points (3-5 items) + - Potential objections and responses + - De-escalation strategies if conversation gets heated + +### 5. **Build Psychological Readiness** + - Ground the user in their intent (why this matters) + - Address anxieties or self-doubt + - Provide emotional regulation techniques + - Affirm their capability to handle this + +### 6. **Create Preparation Document** + - Synthesize above into structured prep guide + - Include what to say, what to avoid, and how to respond + +--- + +## Format + +Structure your output as a **Conversation Preparation Guide** with these sections: + +### 1. Situation Summary (100 words) +Brief restatement of the core issue and desired outcome + +### 2. Emotional Preparation (150 words) +- Your emotions about this conversation +- Their likely emotional state +- Grounding statement (why this matters) + +### 3. Strategic Approach (200 words) +- Clear objective +- Key points to communicate (3-5 bullets) +- Suggested structure (Opening → Middle → Close) + +### 4. Opening Statement (50 words) +Suggested first 30 seconds of the conversation + +### 5. Talking Points & Responses (300 words) +Table format: + +| Your Point | Potential Objection | Your Response | +|-----------|---------------------|---------------| +| [Key point] | [Likely pushback] | [How to respond] | + +### 6. De-escalation Toolkit (150 words) +- If they get defensive... +- If emotions run high... +- If conversation stalls... + +### 7. Self-Care Reminder (50 words) +Brief note on post-conversation processing + +**Total length:** 800-1000 words + +--- + +## Target & Tone + +**Target:** Managers and team leads (3-10 years experience, responsible for difficult conversations but may lack formal training, need both confidence and practical guidance) + +**Characteristics:** +- Intellectually capable but emotionally anxious about conflict +- Want to do right by their team members +- Need practical tools, not just theory +- Time-constrained (15-30 min to prepare) + +**Tone:** +- Empathetic but strategic — validate difficulty while providing structure +- Use "you" language to personalize guidance +- Be direct about hard truths ("This will be uncomfortable, and that's okay") +- Avoid corporate jargon or HR-speak +- Balance emotional support with tactical advice +- Confident and grounding (like a trusted coach in your corner) + +**Communication approach:** +- Lead with validation ("This is genuinely hard") +- Use short paragraphs for scannability +- Include both "what to do" and "why it works" +- Provide examples without being prescriptive +- Empower the user's own judgment + +--- + +## Examples + +### Example 1: Delivering Critical Feedback + +**User input:** +> "I need to tell my team member that their work quality has dropped significantly over the last 2 months. They're defensive when given feedback. I'm worried they'll get upset or quit." + +**Output:** + +```markdown +## Conversation Preparation Guide + +### Situation Summary +You need to address a performance decline with a team member who historically reacts defensively to feedback. Your goal is to understand what's happening, clarify expectations, and create an improvement plan—while maintaining the relationship and their engagement. + +### Emotional Preparation + +**Your emotions:** Anxiety about their reaction, fear of damaging the relationship, maybe frustration about the performance issues. + +**Their likely state:** Possibly stressed or overwhelmed (performance drops often signal something going on). May feel shame if they're aware of the decline. Defensiveness is often a protection mechanism. + +**Grounding statement:** You're having this conversation because you believe in their capability and want them to succeed. Avoiding it would be a disservice to both of them and the team. + +### Strategic Approach + +**Objective:** Open a dialogue to understand what's happening, clarify expectations, and co-create an improvement path. + +**Key points to communicate:** +1. Specific examples of the performance gap (not vague criticism) +2. Impact on the team and projects (why it matters) +3. Your support and belief in their capability +4. Clear expectations going forward +5. Resources or support you can provide + +**Structure:** +- **Opening (2 min):** State purpose, clarify intent ("I'm bringing this up because I care about your success") +- **Middle (15-20 min):** Share observations, ask questions, listen deeply, co-create action plan +- **Close (3 min):** Summarize agreements, reaffirm support, schedule follow-up + +### Opening Statement + +"I wanted to talk with you about some patterns I've noticed in your work over the last couple of months. I'm bringing this up because I value your contribution to the team and want to understand what's going on so we can work through it together. Can we talk about this?" + +### Talking Points & Responses + +| Your Point | Potential Objection | Your Response | +|-----------|---------------------|---------------| +| "The X project had Y quality issues" (be specific) | "I've been really busy" | "I understand—let's talk about workload. What's feeling unsustainable?" | +| "This impacts the team because..." | "No one told me it was a problem" | "You're right, I should have raised this sooner. That's on me. Let's address it now." | +| "I want to understand what's happening for you" | "Everything's fine" (deflection) | "I hear you, and I'm seeing a pattern that concerns me. Help me understand your perspective." | +| "Let's create a plan together" | "I don't know what you want from me" | "Fair. Let me be specific about expectations: [detail 2-3 clear standards]" | + +### De-escalation Toolkit + +**If they get defensive:** +- Pause. Take a breath. Lower your voice slightly. +- "I'm not here to attack you. I'm here because I believe in your capability." +- Shift to questions: "What's your perspective on this?" + +**If emotions run high (theirs or yours):** +- "I can see this is landing hard. Do you need a minute?" +- "Let's take a brief break and come back in 10 minutes." +- Stay calm. Your regulation helps them regulate. + +**If conversation stalls:** +- "What would help right now?" +- "What's one thing we could agree on?" +- Return to shared goal: "We both want you to succeed here." + +### Self-Care Reminder + +After the conversation, take 10 minutes to decompress. This was emotionally demanding. Process with a peer or your own manager if needed. Follow up with your team member within 48 hours (even just a brief check-in). You did a hard thing. That matters. + +--- + +Framework: CoachSteff's CRAFTER (SuperPrompt Framework v0.2) +License: CC-BY 4.0 — Attribution: Steff Vanhaverbeke (coachsteff.live) +``` + +--- + +### Example 2: Addressing Interpersonal Conflict + +**User input:** +> "Two of my team members can't work together anymore. They're professional in meetings but avoid each other and it's creating silos. I need to mediate." + +**Output:** + +```markdown +## Conversation Preparation Guide + +### Situation Summary +You're mediating a conflict between two team members whose working relationship has broken down. They're avoiding each other, creating inefficiency and team strain. Your goal is to facilitate a reset conversation that gets them to a functional working relationship (not friendship—that's not required). + +### Emotional Preparation + +**Your emotions:** Probably frustration ("Why can't they just work it out?"), anxiety about making it worse, maybe irritation at having to manage interpersonal drama. + +**Their likely state:** Both probably feeling justified in their position, possibly resentful, defensive, maybe embarrassed that it's escalated to this point. + +**Grounding statement:** Your job is to create conditions for collaboration, not to judge who's "right." You're the container for a difficult conversation, not the referee. + +### Strategic Approach + +**Objective:** Get both team members to agree on a minimal working relationship—clear communication, respectful interaction, collaborative problem-solving on shared work. + +**Key points:** +1. The current situation is not sustainable (impact on team) +2. You expect professional working relationship (not friendship) +3. Each person needs to hear the other's perspective +4. Focus on future behavior, not past blame +5. Clear agreements about how they'll interact going forward + +**Structure:** +- **Pre-meeting (Individual):** 15 min with each person separately to understand their view +- **Joint meeting (60 min):** Opening → Each shares perspective (uninterrupted) → Identify common ground → Create agreements +- **Follow-up (1 week):** Check-in to see if agreements are holding + +### Opening Statement (Joint Meeting) + +"Thank you both for making time for this. I've noticed the working relationship between you two has broken down, and it's affecting the team. I'm not here to judge who's right or wrong. I'm here to help you both get to a functional working relationship. That doesn't mean you need to be friends, but I do need you to communicate clearly, treat each other with respect, and collaborate effectively on shared projects. Sound fair?" + +### Talking Points & Responses + +| Your Point | Potential Objection | Your Response | +|-----------|---------------------|---------------| +| "Let's hear each person's perspective without interrupting" | [Person interrupts with defense] | "Hold on—you'll have your turn. Right now, I need you to listen." | +| "What's one thing you each appreciate about working with the other?" | "I can't think of anything" | "Try. There must be something, even small. You're both here for a reason." | +| "What do you each need from the other going forward?" | "I need them to change everything" (unrealistic) | "Let's focus on 2-3 specific behaviors that would make the biggest difference." | +| "Let's agree on ground rules for working together" | "This won't work" (defeatist) | "Maybe not perfectly, but we can make it better than it is now. What's one thing you can commit to?" | + +### De-escalation Toolkit + +**If defensiveness escalates:** +- "Let's pause. I'm noticing this is getting heated. Take 30 seconds to breathe." +- Redirect to future: "What happened happened. Let's focus on what you each need going forward." + +**If one dominates:** +- "I appreciate your perspective. Now I need to hear from [other person]." +- Use hand gestures to physically signal turn-taking + +**If it's going nowhere:** +- "We've been at this for 45 minutes. What's the smallest agreement we can make today to move forward?" +- Lower expectations: "You don't have to like each other. You do have to be able to send an email without it taking 3 days." + +### Self-Care Reminder + +Mediation is exhausting. You held space for conflict and facilitated a hard conversation. That's skilled leadership. Don't expect a perfect resolution immediately—you're looking for commitment to try, not transformation. Give it a week, then follow up. If it's not improving, escalate to HR. + +--- + +Framework: CoachSteff's CRAFTER (SuperPrompt Framework v0.2) +License: CC-BY 4.0 — Attribution: Steff Vanhaverbeke (coachsteff.live) +``` + +--- + +## Refining + +**If the user requests changes:** + +### "Make it shorter" +Condense to 3 sections only: +1. Opening Statement (50 words) +2. Key Points (3 bullets, 100 words total) +3. De-escalation (if conversation goes sideways) + +Total: 300 words max + +### "Focus on [specific conversation type]" +Narrow the guidance to their scenario: +- Feedback → Emphasize specificity, evidence, forward focus +- Conflict → Emphasize listening, common ground, agreements +- Termination → Emphasize clarity, dignity, logistics +- Negotiation → Emphasize interests vs positions, creative options + +### "Add more emotional prep" +Expand the Emotional Preparation section with: +- Somatic grounding techniques (box breathing, body scan) +- Cognitive reframing exercises +- Confidence-building affirmations +- Worst-case scenario planning (and why it's unlikely) + +### "Make it more tactical / less emotional" +Reduce emotional validation language, increase: +- Step-by-step scripts +- Decision trees (if they say X, you say Y) +- Behavioral checklists +- Metrics for success + +--- + +Framework: CoachSteff's CRAFTER (SuperPrompt Framework v0.2) +Contributors: [list of contributors] +Pattern Used: [pattern name if applicable - see patterns.md] +License: CC-BY 4.0 — Attribution: Steff Vanhaverbeke (coachsteff.live) + +--- + +## Key Learning: Building from Scratch + +### Design Decisions + +**Why this structure?** +- Managers need both tactical and emotional support for difficult conversations +- Separating sections (strategic vs emotional vs tactical) helps them navigate based on their needs +- Table format for talking points makes it scannable under stress +- De-escalation toolkit addresses common fear ("What if it goes wrong?") + +**Why this tone?** +- Target audience (managers) often feel alone or underprepared for these conversations +- Empathetic tone reduces anxiety, strategic tone builds confidence +- Direct language ("This is hard") validates without being patronizing +- "You" language makes it personal and actionable + +**Why these examples?** +- Feedback and conflict are the two most common difficult conversations +- Showing both demonstrates versatility of the framework +- Each example is concrete enough to be useful but generic enough to adapt + +### Self-Test Process + +Before finalizing, I verified: +- ✅ All 7 CRAFTER components present +- ✅ Target & Tone integrated (managers + empathetic-strategic approach) +- ✅ Actions are specific and numbered +- ✅ Format is clear (section by section with word counts) +- ✅ Examples show complete input→output +- ✅ Refining guidance anticipates common requests + +--- + +## When to Use This Pattern + +Use Mode B (Superprompt Creation) when: + +- ✅ Starting from scratch with a new use case +- ✅ Building reusable prompts for specific workflows +- ✅ Codifying expert knowledge into prompt form +- ✅ Creating tools for teams to use repeatedly +- ✅ Designing prompts that need to work across multiple AI platforms + +**Don't use this pattern when:** +- ❌ You already have a working prompt (use Mode A: Enhancement) +- ❌ The use case is too broad or undefined (clarify first) + +--- + +## Reusability + +This superprompt can be adapted for: +- Different conversation types (negotiation, termination, etc.) +- Different industries (healthcare, education, etc.) +- Different experience levels (new managers vs senior leaders) +- Different cultural contexts (adjust tone for different workplace norms) + +**To adapt:** Replace specifics in Context, adjust Tone in T component, modify Examples to match your domain. + +--- + +**Next Steps:** + +1. Try creating your own superprompt using this pattern +2. Use the self-test checklist (7/7 required) +3. Test your superprompt with an AI model +4. Refine based on output quality +5. Share your superprompt in the [repository](https://github.com/CoachSteff/superprompt-framework) + +--- + +**Questions or feedback?** +Repository: https://github.com/CoachSteff/superprompt-framework +Author: Steff Vanhaverbeke ([@CoachSteff](https://github.com/CoachSteff)) diff --git a/package.json b/package.json index 382ba57..cb7d640 100644 --- a/package.json +++ b/package.json @@ -1,7 +1,7 @@ { "name": "superprompt-framework", - "version": "1.0.0", - "description": "A comprehensive framework for AI prompt engineering and context management", + "version": "1.1.0", + "description": "CoachSteff's CRAFTER SuperPrompt Framework v0.2 - A comprehensive framework for AI prompt engineering and context management", "main": "index.js", "scripts": { "test": "echo \"Error: no test specified\" && exit 1",