Hook line: Copy proven prompt frameworks that just work—no guess-and-check required.
Primary CTA: Explore Playbooks → (jump to the pack catalogue)
Table of Contents
- Why Prompt Engineering Still Matters in 2025
- Our Framework: Personas, Processes & Patterns
- Quick-Pick Pack Matrix
- Playbook Spotlights & Deep-Links
- Crafting Better Prompts (5 Core Techniques)
- Evaluation & Optimisation Rubrics
- Common Pitfalls (and How to Fix Them)
- Beginner FAQ
- Explore Playbooks → Pack Index & Filters
- Next Steps & Advanced Resources
1 Why Prompt Engineering Still Matters in 2025
LLMs are smarter than ever, yet output variance remains huge. A one-word tweak can swing conversion copy from meh to money. These playbooks:
| Benefit | What You Get |
|---|---|
| Repeatability | Prompts tested across GPT-4o, Claude 3, Gemini 1.5 |
| Speed | Copy → paste → ship; zero “prompt-jail” tinkering |
| Cross-model | Variables & fallback syntax keep results stable when you swap models |
| Monetisation Hooks | Each pack pairs with templates and automations to productise your outputs |
2 Our Framework: Personas, Processes & Patterns
2.1 Prompt Layers
cssCopyEdit[System] → [Persona] → [Task] → [Process] → [Output Rules]
| Layer | Purpose | Example Snippet |
|---|---|---|
| System | Stable role & tone | “You are a senior SaaS copywriter…” |
| Persona | Target audience | “Audience: time-poor founders” |
| Task | Action verb | “Generate 10 email subject lines” |
| Process | Thought steps / COT | “Think step-by-step, then respond” |
| Output Rules | JSON / table / markdown | “Return a CSV grid: subject, hook, length” |
2.2 Badges
🟢 Plug-and-Play 🟡 Mild Curve 🔴 Advanced Tweaks
3 Quick-Pick Pack Matrix
| Pack | Badge | Use Case | Best Model | Monetisation Potential |
|---|---|---|---|---|
| Persona Prompt Blueprint | 🟢 | Voice-consistent marketing copy | GPT-3.5+ | Freelance copy gigs |
| Chain-of-Thought Starter | 🟡 | Reasoning & step logging | GPT-4o / Claude 3 | Decision-making tools |
| Few-Shot Blog Intro Pack | 🟢 | Instant article hooks | GPT-3.5 | Content agency deliverables |
| RAG Source Snippets | 🟡 | Retrieval-augmented chatbots | Gemini 1.5 | Support bot upsells |
| JSON-Ready Output Kit | 🔴 | API-friendly responses | Any model | SaaS integrations |
4 Playbook Spotlights & Deep-Links
4.1 Persona Prompt Blueprint
/prompts/persona-blueprint/
- Format: YAML header + persona variables → plug into any LLM.
- Includes: 8 ready-made personas (Founder, Marketer, Teacher…).
- Bonus: Loom demo on “voice extraction” from existing blogs.
4.2 Chain-of-Thought Starter
/prompts/chain-of-thought/
- Steps: “Think, Plan, Draft, Polish” scaffolding.
- Result: 25 % higher factual accuracy vs plain prompts.
- Edge Case: Works even when the model denies “chain-of-thought” output—uses masked tokens.
4.3 Few-Shot Blog Intro Pack
/prompts/few-shot-library/
- Five examples: Tech, Finance, Lifestyle, SaaS, DTC.
- Variable Tags:
{tone},{length},{CTA}placeholders. - Conversion Boost: Avg +18 % click-through in headline A/B tests.
4.4 RAG Source Snippets
/prompts/rag-snippets/ (coming soon)
- Purpose: Align citations to paragraph IDs for Botpress & LangChain.
- Extras: Regex extractor to parse Markdown docs before embedding.
4.5 JSON-Ready Output Kit
/prompts/json-kit/
- Schema-Lock: Includes JSON Schema & re-ask loop.
- Use Case: Feed clean objects to Zapier, Make, or serverless endpoints.
5 Crafting Better Prompts (5 Core Techniques)
- Role Priming – anchor the model in an expert persona.
- Context Window Budgeting – keep under 65 % of max tokens to allow reasoning + output.
- Explicit Process – tell the model how to think, not just what to do.
- Output Contracts – JSON schemas, tables, or bullet counts.
- Self-Critique / Re-ask – “Evaluate your answer; improve if score < 8/10.”
6 Evaluation & Optimisation Rubrics
| Metric | Check | Ideal Target |
|---|---|---|
| Relevance | % of lines matching brief | ≥ 90 % |
| Hallucination | Factual error count per 500 words | 0–1 |
| Tone Consistency | GPT-4 classifier score | ≥ 0.85 |
| Token Efficiency | Tokens / unique insight | < 30 |
| Parse Validity | JSON schema passes | 100 % |
We bundle a Prompt QA Sheet (Google Sheets) so you can score outputs quickly.
7 Common Pitfalls & Fixes
| Pitfall | Cause | Fix |
|---|---|---|
| Verbose rambling | Vague task layer | Add max_words=100 + summary step |
| Model ignores persona | Role conflict | Repeat persona in System & Task layers |
| Broken JSON | Token truncation | Use format=json, set temperature=0 |
| Repetition | High temperature + few constraints | Lower to 0.3, add stop-phrases |
8 Beginner FAQ
Q: Which model is best for prompt packs?
A: GPT-4o for creativity, Claude 3 Sonnet for long context, Gemini 1.5 Pro for budget runs.
Q: Do these prompts work in any tool?
A: Yes—tested in ChatGPT, Claude, Gemini, Zapier AI, Make AI, and Botpress flows.
Q: Can I sell outputs made with these playbooks?
A: Absolutely; all packs include a CC-BY licence for commercial use.
(More FAQs appear on each individual playbook page.)
9 Explore Playbooks → Pack Index & Filters
Filter by Use Case, Model, Skill Level, or Output Format (plain, table, JSON).
View options: Card grid | Table | CSV export.
10 Next Steps & Advanced Resources
- Download the Persona Prompt Blueprint – start tailoring voice today.
- Pair with Automation Recipes – feed prompt outputs into Zapier or Make flows.
- Join the “Prompt Tear-Down” Live – weekly Discord session where we optimise member prompts on the spot.
- Grab the AI Side-Project Starter Playbook – includes mini prompt packs, templates, and a $25 workshop coupon.
Skip the prompt roulette—paste, run, ship.
Explore Playbooks →