"Prompt Engineering 101 (2026) — Five Patterns for Asking Better Questions"

Distilled from Anthropic and OpenAI's official guides, with copy-paste templates anyone can use

ํ•ต์‹ฌ ์š”์•ฝ

  • Audience: Anyone using ChatGPT, Claude, or Gemini who keeps getting answers that miss the point — plus developers writing their first prompts.
  • What you'll get: 1) Five patterns for better prompts (per Anthropic/OpenAI official docs), 2) copy-paste templates for each, 3) the seven most common mistakes, 4) a before/after example, 5) what 2026 research changed.
  • Mental model: An LLM isn't a wizard. It's a brand-new intern. Brief it the way you'd brief a smart human who has zero context.

1. Why the same question gives different answers

LLMs continue your input statistically. Vague prompt → vague answer. Specific prompt → specific answer. "Write something good" and "Write a 50-character SNS ad headline targeting 30-something professionals, casual tone, one emoji" share intent — but the output gap is enormous.

Anthropic's official guide lists five core techniques: be clear and direct, give examples, let the model think step by step, assign a role, use structure (XML tags). The five patterns below are exactly those, translated for everyday users.

2. Pattern 1 — Be clear and direct

Bad

"Write a marketing post."

Good

"Write an SNS ad caption for newly-released wireless earbuds, targeted at women in their 30s who work full-time. Highlight three core values. One sentence under 50 characters. Include one emoji. Friendly but professional tone."

Checklist

  • [ ] Audience: Who's reading? (women in their 30s, full-time workers)
  • [ ] Format: What kind of output? (SNS ad caption)
  • [ ] Constraints: Length, count, structure (50 chars, 1 emoji)
  • [ ] Style: Tone, voice, register (friendly-professional)

If retyping this every time is annoying, save it once: ChatGPT Custom GPTs, Claude Projects, Gemini Gems.

3. Pattern 2 — Show examples (few-shot)

A few examples beat paragraphs of explanation.

Bad

"Turn my journal entry into a tweet."

Good

"Convert journal entries into tweets in this style.

Example 1. Journal: "Had ramen for lunch. 30-minute line. Worth it." Tweet: "30 minutes in line for ramen. One bite forgave everything ๐Ÿœ"

Example 2. Journal: "Saw a movie with a friend on the weekend. Moving." Tweet: "Old friend + unexpected masterpiece. Wish weekends were longer."

Now convert: "Worked until dawn. Looked up — already 6 AM.""

Why it works

2026 research shows 1–2 examples deliver an average 7.48% accuracy lift over zero-shot, with diminishing returns past 4–5 examples (Few-Shot Prompting 2026 guide). 2–3 diverse examples is the sweet spot.

4. Pattern 3 — Let it think (chain of thought)

For anything involving reasoning or calculation, adding "think step by step" raises accuracy with almost zero effort.

Bad

"Is this remodeling quote reasonable? (quote attached)"

Good

"Decide whether this remodeling quote is reasonable.

Walk through the analysis in this order: 1. Compare each line item against market rates 2. Flag missing or duplicated items 3. Check whether material grade matches the price 4. Final verdict (fair / overpriced / underbid + why)

Quote: ..."

What changed in 2026

Recent research suggests strong models (GPT-5, Claude 4) usually need only "think step by step" — not full chain-of-thought examples (Zero-shot CoT paper). Add demonstrations only for hard problems or weaker models.

5. Pattern 4 — Assign a role

"You are [role]" upfront aligns tone, perspective, and expertise in one shot.

Bad

"Review this contract."

Good

"You are a contract lawyer specializing in IT startups. Review the SaaS agreement below and produce: 1. Clauses that are risky from the supplier (us) perspective 2. Phrasing that's ambiguous or could be read multiple ways 3. Protective clauses we should add

Output in plain English, bulletized."

Cautions

  • Be specific: "lawyer" < "IT startup contract lawyer."
  • Roles don't fix knowledge gaps. A "doctor" persona still hallucinates if the underlying training didn't cover it.
  • Avoid impersonating real people / celebrities — ethics + IP issues.

6. Pattern 5 — Add structure (XML tags or sections)

Long inputs need separators. Without them, the model can confuse instructions with content.

Bad

"Use the document below to draft a reply to this email." (everything pasted as plain text)

Good

``` [Full company refund policy]

[Original customer email]

Draft a reply that respects the policy. Polite but firm. ```

XML tags are Anthropic's recommendation. Markdown headers (### Context, ### Email, ### Task) work fine for OpenAI. Pick one and use it consistently.

7. The seven most common mistakes

Mistake Result Fix
1. Cramming many requests into one prompt Shallow or partial answers One prompt = one task
2. Negative instructions ("don't do X") Often ignored Reframe as positive ("do Y instead")
3. Vague qualifiers ("make it nice", "well") Inconsistent output Measurable criteria ("3 bullets," "under 300 words")
4. No context, just a directive Generic answer Lead with 1–2 lines of background and goal
5. Stacking unrelated questions in one chat Earlier turns bias the model New chat per topic
6. "Be accurate" Confident guesses "If unsure, reply 'uncertain'"
7. Restarting from scratch when unsatisfied Wasted tokens "From the previous answer, change only X"

8. Before / After — one more

Before

"Make me a presentation."

After

"You are a B2B SaaS sales coach.

Build a 30-minute pitch outline for a sales meeting.

Audience: a 50-person IT company — 1 CTO, 2 engineering leads. Our product: a cloud security automation SaaS at $500/month. Goal: book a follow-up PoC meeting after the 30 minutes.

Output as a slide-by-slide outline: - Slide number - Slide title (under 10 words) - Core message (one line) - Audience-engagement moment (if any)

8–12 slides total."

The wins are specificity and forced output structure. Same model, same cost, ~5× quality difference.

Developer notes

For API users:

  1. System vs. user prompts: Persona, rules, persistent constraints → system. Per-request input → user. Better for caching too.
  2. XML tags are more reliable than Markdown in mixed-content prompts. Claude 4 series is especially good at XML. OpenAI handles either fine.
  3. Few examples, but diverse: Two distinct cases beats five near-duplicates.
  4. Force JSON Schema when format matters: OpenAI response_format, Anthropic tool_use with a strict schema.
  5. Temperature: 0–0.3 for accuracy-critical tasks, 0.7–1.0 for creative work.
  6. Anthropic's interactive tutorial — free 9-chapter course. Worth one full pass if you're going deep.

References

This is part 3 of 11 in the AI Basics series. Next: Advanced prompting — roles, examples, and chaining in practice (extension).

๋Œ“๊ธ€

๋Œ“๊ธ€ ์“ฐ๊ธฐ

์ด ๋ธ”๋กœ๊ทธ์˜ ์ธ๊ธฐ ๊ฒŒ์‹œ๋ฌผ

Agent Memory Engine (2/10) — Building an AI Agent Memory System with SQLite Alone

← 1/10 SQLite Memory Engine f… ๐Ÿ“š Series Index 3/10 memcore Completed → A memory engine that runs without daemons, without dependencies, anywhere ํ•ต์‹ฌ ์š”์•ฝ This post covers how to design an AI agent memory engine (memcore) on a single-file SQLite backend. - Structure: 25 modules, ~6,464 lines, 22 DB tables - Principles: no daemon, single-file portability, no required external dependencies, host-agnostic - Core techniques: 4-layer hybrid search (topic · vector · FTS5 · LIKE), U-tag dialectic-based confidence evolution, ingestion-layer bias rejection rules What This Post Covers One approach to building AI agent memory without a server or cloud vector DB. Specifically: (1) the limitations of text-file-based memory, (2) vector DB alternatives compared and selection criteria, (3) how to layer search tiers, (4) how to evolve confidence scores over time, and (5) how to block biased memory accumulation at the ingestion stage. Design Principles: Why Single-File SQLite Text-file-based ...
Read more »

"ML Foundations (9/9) — PyTorch vs TensorFlow, and the Road to Local LLMs"

์ด๋ฏธ์ง€
← 8/9 Deep Learning Architec… ๐Ÿ“š Series Index (series end) The final part. So far we've focused on models. Now we focus on the tools that actually run them . We'll lay out the philosophical difference between PyTorch and TensorFlow, and trace how Hugging Face transformers , llama.cpp , MLX , and Ollama built the bridge to running large language models on your own machine. By the end you should have the full mental model of "download a pretrained LLM and serve it locally." 0. Learning Objectives Explain the eager-vs-graph difference between PyTorch and TensorFlow. Explain, in graph terms, how autograd automates the backward pass. Use Hugging Face transformers ' AutoModel , AutoTokenizer , and pipeline abstractions. Describe llama.cpp's GGUF quantization, INT4 inference, and CPU-first flow. Describe Apple MLX's use of unified memory and how it differs from PyTorch. Run a local LLM with Ollama and call it through an OpenAI-compatible API. ...
Read more »

"RAG Core Study (14/26) — Evaluation Sets with RAGAS & DeepEval"

์ด๋ฏธ์ง€
Series overview: Series index ๐Ÿ“š Series Index Without an evaluation set, every RAG improvement is a story, not evidence. RAG systems fail in more than one place: retrieval, context selection, answer generation, and grounding. That means "looks better" is not a metric. Part 14 explains how to build a golden dataset , how RAGAS and DeepEval fit into the workflow, and why teams must separate retrieval quality from answer quality if they want tuning decisions to hold up. 0. Prerequisites Part 13 reranking — system quality now depends on multi-stage ranking. Part 1 grounding — a correct-looking answer is not enough. Part 12 Hybrid — multiple retrieval settings need comparison on fixed data. 1. Learning Objectives Build a small but useful golden dataset for RAG. Distinguish retrieval metrics from answer metrics . Use RAGAS and DeepEval in the right roles. Avoid the common traps in judge-model-based evaluation. 2. ํ•ต์‹ฌ ์š”์•ฝ An evaluation set for RAG is ...
Read more »

"ML Foundations (8/9) — Deep Learning Architectures: CNN, RNN, Attention"

์ด๋ฏธ์ง€
← 7/9 Deep Learning Training ๐Ÿ“š Series Index 9/9 PyTorch vs TensorFlow → What did we keep adding on top of an MLP? Vision went CNN, sequences went RNN/LSTM, and eventually both converged on Attention and the Transformer. Each architecture is easier to remember if you read it as what it gave up to gain something else . This part is the one-line summary of the decisive inflection points: LeNet → AlexNet → ResNet → LSTM → Attention → Transformer. 0. Learning Objectives Explain why CNN's convolution, pooling, and weight sharing match images so well. Trace what got unblocked from LeNet → AlexNet → VGG → ResNet as depth grew. State BPTT for RNNs and the vanishing/exploding gradient problem. Write the LSTM gate equations (forget, input, output) with intuition. Write the attention formula in query/key/value form and the scaled dot-product variant. State the precise reasons Transformers replaced RNNs (parallelism + long-range dependencies). 1. ํ•ต์‹ฌ ์š”์•ฝ CNN : weight sharin...
Read more »

"ML Foundations (7/9) — Deep Learning Training: Optimizers, Regularization, Initialization"

← 6/9 Neural Networks ๐Ÿ“š Series Index 8/9 Deep Learning Architec… → Building an MLP, as we did in Part 6, is not the same as getting it to train well . The right optimizer, regularization, initialization, and learning rate — when these line up, deep networks converge. When they don't, the network refuses to learn at all. This part is the catalog of those crafts, with formulas, paper citations, and code in one place. 0. Learning Objectives Compare and write the update rules for SGD, Momentum, Nesterov, and Adam. Explain how Dropout, BatchNorm, and LayerNorm work and where they belong in a model. Derive the variance formulas for Xavier and He initialization and match them to activations. Implement step, cosine, and warmup learning-rate schedules in PyTorch. Explain why gradient clipping is effectively required for RNNs and Transformers. Diagnose the most common training failures (NaN, plateau, overfitting) and apply first-line fixes. 1. ํ•ต์‹ฌ ์š”์•ฝ SGD : \(w \leftarro...
Read more »

OpenClaw to Hermes Migration (2/13) — What to Preserve, Partially Port, or Discard

← 1/13 Current Structure Inve… ๐Ÿ“š Series Index 3/13 What Moves to Hermes → A code review record classifying operational scripts into three tiers: preserve, partial-preserve, and discard. What This Post Covers An asset classification framework (preserve / partial-preserve / discard) for migrating a legacy agent system to a successor A concrete case applying the Strangler Fig + Subset Migration strategy to a memory pipeline Classification results evaluated across LOC, external dependencies, and replaceability axes A method for slimming down a bloated single file (1,310 LOC) using the migration as a natural entry point Problem Definition: Port Everything or Cut Deep? The OpenClaw inventory comprises multiple agents, multiple scheduler/daemon processes, and multiple operational scripts. When moving to the successor system, Hermes, the choice is not binary. A full port carries accumulated technical debt along with it; a minimal port discards hard-won domain logic. Code-revie...
Read more »

AI Agents I Built (5/7) — Building an Automated Blogger API Publishing System

← 4/7 My Life Organizer Agent ๐Ÿ“š Series Index 6/7 Dual → Designing a Markdown-Based Auto-Publish Pipeline with Blogger API v3 + OAuth 2.0 Key Summary Blogger API v3 + OAuth 2.0 enables a pipeline that converts Markdown files into publishable HTML posts blogger_publish.py handles the full flow — frontmatter parsing → body cleanup → HTML conversion → CSS injection → API post — in a single script For bulk publishing, quota management (delay, retry, TOC suppression) is the critical design concern Platform Selection Criteria API access is mandatory for AI agent-driven content publishing. A comparison of platforms: Platform API Cost AdSense Decision WordPress.com REST API Paid plan required Paid only Cost overhead Ghost REST API Self-hosted or paid Manual setup Ops overhead Blogger v3 API Free Native integration Adopted Blogger is free, exposes a REST API, and integrates natively with AdSense. For personal blog automation, this combination i...
Read more »