Prompt Quest

Tracks

1. 01✍️Prompting Foundations

   0/10 lessons

   What a prompt is, what it is not, and why modern models changed the job

   Build the mental model: prompts are product contracts for probabilistic systems, not lucky phrases. We'll fix the most common mistakes before they cost you anything.

   Lesson list (10)

   • 01
     A Prompt Is a Contract, Not a Spell~18 min · foundations, contract, intent
   • 02
     Frontier Models Changed What Prompts Are For~20 min · foundations, frontier-models, history
   • 03
     The Five Layers of a Prompt Contract~22 min · foundations, structure, anatomy
   • 04
     Specificity Beats Length~16 min · foundations, specificity, concision
   • 05
     Probabilistic Output Means Test, Don't Hope~18 min · foundations, evaluation, non-determinism
   • 06
     The Cost of Bad Prompts in Production~16 min · foundations, production, cost
   • 07
     Prompts as Versioned Code, Not Strings~18 min · foundations, version-control, engineering
   • 08
     Reading the Sampler — Temperature, Top-P, and Seeds~16 min · foundations, sampling, determinism
   • 09
     Speech Act, Not Magic Phrase~14 min · foundations, linguistics, intent
   • 10
     Foundations You'll Hit Every Day~14 min · foundations, review, habits

   ✓Quiz · 4 questions→

2. 02🧱Writing Instructions That Hold

   0/10 lessons

   Imperatives, constraints, examples, and the anti-patterns that quietly ruin them

   Instructions are the directive layer of the contract. This track is the daily craft of writing them so they survive edge cases, edits, and other people's hands.

   Lesson list (10)

   • 01
     Imperative Voice — Write the Verb First~14 min · instructions, voice, clarity
   • 02
     Negative Constraints vs Positive Specifications~16 min · instructions, constraints
   • 03
     Few-Shot Examples — When and How Many~22 min · instructions, examples, few-shot
   • 04
     Role and Persona — Useful Without Being a Cargo Cult~16 min · instructions, persona, role
   • 05
     Constraints That Survive Edge Cases~18 min · instructions, edge-cases, robustness
   • 06
     Step-by-Step vs Open-Ended~14 min · instructions, structure, process
   • 07
     Boundary Conditions — What to Do When Information Is Missing~16 min · instructions, uncertainty, honesty
   • 08
     Output Shape Description Before Examples~14 min · instructions, shape, format
   • 09
     Anti-Patterns — "Be Helpful," "Think Carefully," "Take Your Time"~14 min · instructions, anti-patterns
   • 10
     Editing Drift — Why Instructions Stack Badly~16 min · instructions, maintenance, drift

   ✓Quiz · 4 questions→

3. 03🗂️Packing Context That Helps the Model Help You

   0/10 lessons

   RAG, chunks, caching, ordering, and the tokens you don't need to send

   Modern prompts are mostly context. This track is the engineering of what to put in front of the model, in what order, and what to leave out.

   Lesson list (10)

   • 01
     Context Window Math — Tokens, Order, and What Stays~18 min · context, tokens, math
   • 02
     Retrieval-Augmented Generation Is Not Magic Glue~22 min · context, rag, retrieval
   • 03
     Chunking Strategies That Don't Lose the Argument~18 min · context, chunking, rag
   • 04
     Reranking — When the Top-K Lies~16 min · context, reranking, rag
   • 05
     Citing Sources Inside the Prompt~14 min · context, citations, grounding
   • 06
     Context Caching — How and Where to Use It~18 min · context, caching, cost
   • 07
     Long-Context Limits — When the Model Forgets the Middle~16 min · context, long-context, attention
   • 08
     Context Compression — Summarize What You Don't Need~14 min · context, compression, cost
   • 09
     Multi-Document Prompts — Order, Boundaries, Naming~16 min · context, multi-document, structure
   • 10
     The Prompt Is Smaller Than the Context~12 min · context, prompt-vs-context, mental-model

   ✓Quiz · 4 questions→

4. 04🧠Directing Reasoning Instead of Hoping for It

   0/10 lessons

   CoT, extended thinking, verification, budgets, and the theatre of fake reasoning

   Modern models can think for you — sometimes too much. This track is about wiring the reasoning so it lands where the answer needs it, and not where it doesn't.

   Lesson list (10)

   • 01
     Chain-of-Thought — When It Helps and When It Hurts~18 min · reasoning, cot
   • 02
     Extended Thinking Modes — Claude, OpenAI, Gemini Differences~20 min · reasoning, extended-thinking, providers
   • 03
     Tree-of-Thought and Self-Consistency~16 min · reasoning, tot, self-consistency
   • 04
     Reasoning vs Output — Separating the Two~14 min · reasoning, output, structure
   • 05
     Verifier-Driven Reasoning~18 min · reasoning, verification
   • 06
     Step Compression — Don't Show Your Work When You Don't Have To~12 min · reasoning, compression, cost
   • 07
     Reasoning Budget vs Output Budget~12 min · reasoning, budget, max-tokens
   • 08
     The Trap of Reasoning Theatre~12 min · reasoning, anti-patterns
   • 09
     Calibrating Confidence in Reasoning Output~14 min · reasoning, confidence, calibration
   • 10
     Debugging Bad Reasoning — Read the Trace~14 min · reasoning, debugging

   ✓Quiz · 4 questions→

5. 05🧩Forging Outputs and Tool Calls That the Model Honors

   0/10 lessons

   Structured outputs, schemas, tool choice, multi-tool orchestration, and recovery

   By 2026, prompts no longer end at 'return JSON.' They name schemas, expose tools, choose among them, and recover from failed calls. This track is the structured side of prompting.

   Lesson list (10)

   • 01
     Structured Outputs — JSON Mode and Beyond~18 min · outputs, json, structured
   • 02
     Schema-First Prompting~14 min · outputs, schema, design
   • 03
     Tool Calling — How the Model Decides~18 min · tools, tool-calling
   • 04
     Tool Choice — required vs auto vs none~12 min · tools, tool-choice
   • 05
     Multi-Tool Orchestration~20 min · tools, agents, orchestration
   • 06
     Tool Errors — Letting the Model Recover~14 min · tools, error-handling
   • 07
     Streaming Output — Partial JSON and the Pipeline~14 min · outputs, streaming
   • 08
     Output Repair Loops~14 min · outputs, repair
   • 09
     Pydantic / Zod / Schemas as Contracts~14 min · outputs, schema, contracts
   • 10
     Citations and Traceability in Output~12 min · outputs, citations, traceability

   ✓Quiz · 4 questions→

6. 06🤝Multi-Turn Conversations That Stay on Contract

   0/10 lessons

   System / user / assistant layers, persona drift, compaction, and external memory

   Single-turn prompts are the easy case. Real systems are multi-turn — and that's where state, drift, and refusal patterns live.

   Lesson list (10)

   • 01
     The Multi-Turn Contract~16 min · conversation, contract, multi-turn
   • 02
     System / Developer / User / Assistant Prefill~14 min · conversation, layers, prefill
   • 03
     Memory — When the Model Should Remember and When It Shouldn't~16 min · conversation, memory
   • 04
     Persona Drift — Catching It Across Turns~14 min · conversation, persona, drift
   • 05
     Long Conversations — Compaction Strategies~18 min · conversation, compaction, long-context
   • 06
     Re-Anchoring at Each Turn~12 min · conversation, anchoring
   • 07
     Stateful Tools in a Stateless Model~16 min · conversation, tools, state
   • 08
     The Pippa Pattern — Vault as External Memory~14 min · conversation, memory, self-reference
   • 09
     Refusals That Don't Kill the Conversation~14 min · conversation, refusals
   • 10
     Closing the Loop — When to End the Conversation~12 min · conversation, closing

   ✓Quiz · 4 questions→

7. 07🛡️Defending Prompts Against Real Adversaries

   0/10 lessons

   Prompt injection, jailbreaks, indirect attacks, output filtering, sandboxing

   Prompt injection is the OWASP-top concern of LLM apps. This track is the threat model and the layered defenses that actually work.

   Lesson list (10)

   • 01
     Prompt Injection Is Not a Trick — It's a Threat Model~18 min · security, injection, threat-model
   • 02
     Defense in Depth — Layers, Not Patches~16 min · security, defense-in-depth
   • 03
     Indirect Injection From Tool Outputs~16 min · security, indirect-injection, tools
   • 04
     Jailbreak Patterns — What They Look Like~16 min · security, jailbreaks
   • 05
     Output Filtering — Don't Trust the Model's Words~14 min · security, output-filtering
   • 06
     Privilege Boundaries — What the Model Can and Can't Do~14 min · security, privilege
   • 07
     Untrusted Content Tags — XML, JSON, Markdown~12 min · security, tagging
   • 08
     Sandboxing Tool Calls~14 min · security, sandbox, tools
   • 09
     Audit Trails — When Something Bad Happens~12 min · security, audit, logging
   • 10
     Red Teaming Your Own Prompts~14 min · security, red-team

   ✓Quiz · 4 questions→

8. 08📏Evals That Catch Regressions Before Users Do

   0/10 lessons

   Golden sets, pairwise judging, A/B in prod, metric design, drift

   Evaluation is what turns prompt engineering from craft into engineering. This track is the loop that proves your prompts work — and keeps proving it.

   Lesson list (10)

   • 01
     Evals Before Prompts — Why Working Backward Wins~16 min · evaluation, process
   • 02
     Golden Sets — Building and Maintaining Them~16 min · evaluation, golden-sets
   • 03
     Pairwise Comparison — When Absolute Scoring Lies~14 min · evaluation, pairwise
   • 04
     LLM-as-Judge — Use Cases and Pitfalls~16 min · evaluation, judges
   • 05
     Regression Tests for Prompts~12 min · evaluation, regression
   • 06
     Cost-Quality-Latency Triangles~14 min · evaluation, tradeoffs
   • 07
     Metric Design — Don't Measure What's Easy~14 min · evaluation, metrics
   • 08
     A/B Prompts in Production~14 min · evaluation, ab-testing
   • 09
     Eval Drift — When Your Test Set Lies~12 min · evaluation, drift
   • 10
     Treating Evals as Code~12 min · evaluation, engineering

   ✓Quiz · 4 questions→

9. 09🔀Across Providers Without Losing Intent

   0/10 lessons

   Claude, OpenAI, Gemini, open-source — how the same prompt behaves and breaks differently

   Real systems span providers. This track is the differences that matter when porting prompts and the adapter pattern that keeps the rest of your stack provider-neutral.

   Lesson list (10)

   • 01
     Why Providers Differ — Beyond the Marketing~14 min · providers, comparison
   • 02
     Claude — System Prompts, Long Context, XML Tags~14 min · providers, claude
   • 03
     OpenAI — JSON Mode, Function Calling, Reasoning~14 min · providers, openai
   • 04
     Gemini — Multimodal, Long Context, Reasoning~14 min · providers, gemini
   • 05
     Open-Source Models — Llama, Qwen, Mistral~14 min · providers, oss
   • 06
     Switching Providers — What Breaks and How to Plan~14 min · providers, migration
   • 07
     Modality Differences — Vision, Audio, Files~14 min · providers, multimodal
   • 08
     Pricing and Rate Limits — Real Numbers~12 min · providers, pricing
   • 09
     SDK Quirks — Retries, Streaming, Cancellation~12 min · providers, sdk
   • 10
     The Adapter Pattern for Multi-Provider Code~14 min · providers, adapter

   ✓Quiz · 4 questions→

10. 10🏗️Shipping Prompts to Production With Confidence

    0/10 lessons

    Versioning, caching, logging, fallback chains, rollouts, cost engineering, incidents

    The last track is the operational maturity ladder. By the end, your prompts behave like the production-grade systems they are — versioned, observable, recoverable, accountable.

    Lesson list (10)

    • 01
      Versioning Prompts — Git, Not Strings~14 min · production, versioning
    • 02
      Caching Strategies — What's Worth Caching~12 min · production, caching
    • 03
      Logging Prompts and Responses Without Leaking~12 min · production, logging, privacy
    • 04
      Telemetry — What to Measure in Prod~12 min · production, telemetry
    • 05
      Fallback Chains — Multiple Models, One Promise~14 min · production, fallback
    • 06
      Rate Limiting and Backpressure~12 min · production, rate-limits
    • 07
      Rollouts and Canaries~12 min · production, rollouts
    • 08
      Cost Engineering — Where the Money Goes~14 min · production, cost
    • 09
      Incident Response for AI Systems~14 min · production, incidents
    • 10
      The Production Maturity Ladder~12 min · production, maturity

    ✓Quiz · 4 questions→