# ARCHITECT'S ASCENT ## A 12-Week Quest to Claude Certified Architect -- Foundations ``` /\ / \ You are an apprentice architect. / ?? \ Your quest: ascend the Tower of Claude /______\ and earn the title of Certified Architect. | | | DOOR | Each floor is a domain of knowledge. |______| Each boss is a practice test. /________\ The final boss: the certification exam. ``` --- ## THE RULES **XP System:** Earn experience points through daily study and challenges. **Levels:** Progress from Apprentice to Certified Architect across 12 weeks. **Streaks:** Consecutive days of study multiply your XP. **Boss Battles:** Weekly practice tests determine if you advance to the next floor. ### XP Awards | Activity | XP | |----------|-----| | Daily study session (1 hour) | 100 XP | | Boss Battle passed (7/10+) | 500 XP | | Boss Battle perfect (10/10) | 1000 XP | | Full Exam passed (36/50+) | 2000 XP | | Full Exam perfect (45/50+) | 5000 XP | | Hands-on exercise completed | 300 XP | | Streak bonus (per consecutive day) | +10 XP/day | | Speed Run bonus (boss in <15 min) | 200 XP | | Teaching bonus (explain a concept from memory) | 150 XP | ### Streak Multipliers | Streak | Multiplier | |--------|------------| | 1-6 days | 1x | | 7-13 days (1 week) | 1.5x | | 14-20 days (2 weeks) | 2x | | 21-29 days (3 weeks) | 2.5x | | 30+ days (1 month) | 3x | --- ## CHARACTER PROGRESSION ### Level 1: APPRENTICE (Weeks 1-2) -- 0 XP ``` O Weapon: Basic API Knowledge /|\ Armor: None / \ Special: Can check stop_reason ``` **Quest Line:** The Agentic Loop Trials - Floor 1: The Loop Chamber (agentic loops, stop_reason, tool results) - Floor 2: The Coordination Hall (multi-agent, hub-and-spoke, Task tool) - **Boss 1:** Test 01 -- Agentic Loops (10 questions) - **Boss 2:** Test 02 -- Multi-Agent Systems (10 questions) - **Unlock at 1000 XP:** Title "Loop Warden" ### Level 2: JOURNEYMAN (Weeks 3-4) -- 2000 XP ``` O Weapon: Programmatic Hooks /|\ Armor: Structured Errors / \ Special: Can intercept tool calls ``` **Quest Line:** The Hook Forge & Tool Armory - Floor 3: The Hook Forge (PostToolUse, interception, enforcement vs guidance) - Floor 4: The Tool Armory (tool descriptions, MCP, error responses) - **Boss 3:** Test 03 -- Hooks & Workflows (10 questions) - **Boss 4:** Test 04 -- Tool Design & MCP (10 questions) - **Unlock at 4000 XP:** Title "Hook Smith" ### Level 3: BUILDER (Weeks 5-6) -- 5000 XP ``` O Weapon: CLAUDE.md Hierarchy /|\ Armor: Path-Scoped Rules / \ Special: Can configure CI pipelines ``` **Quest Line:** The Configuration Labyrinth - Floor 5: The Config Maze (CLAUDE.md hierarchy, rules, skills, commands) - Floor 6: The Pipeline Forge (plan mode, CI/CD, -p flag, batch API) - **Boss 5:** Test 05 -- Claude Code Config (10 questions) - **Boss 6:** Test 06 -- Plan Mode & CI/CD (10 questions) - **Unlock at 8000 XP:** Title "Config Architect" ### Level 4: ARCHITECT (Weeks 7-8) -- 9000 XP ``` O Weapon: JSON Schemas /|\ Armor: Few-Shot Examples / \ Special: Can force tool_choice ``` **Quest Line:** The Schema Sanctum - Floor 7: The Prompt Workshop (explicit criteria, few-shot, tool_use) - Floor 8: The Validation Gauntlet (retry loops, batch, multi-pass review) - **Boss 7:** Test 07 -- Prompt Engineering (10 questions) - **Boss 8:** Test 08 -- Validation & Multi-Pass (10 questions) - **Unlock at 13000 XP:** Title "Schema Sage" ### Level 5: MASTER ARCHITECT (Weeks 9-10) -- 14000 XP ``` O Weapon: Context Management /|\ Armor: Provenance Tracking / \ Special: Can manage multi-source synthesis ``` **Quest Line:** The Context Tower - Floor 9: The Memory Halls (context management, escalation, error propagation) - Floor 10: The Synthesis Chamber (provenance, temporal data, conflict annotation) - **Boss 9:** Test 09 -- Context & Reliability (10 questions) - **Boss 10:** Test 10 -- Advanced Context (10 questions) - **Unlock at 18000 XP:** Title "Context Keeper" ### Level 6: CERTIFIED ARCHITECT (Weeks 11-12) -- 20000 XP ``` O Weapon: Full Arsenal /|\ Armor: Complete Knowledge / \ Special: All abilities unlocked ``` **Quest Line:** The Final Ascent - The Proving Grounds (4 hands-on exercises) - **FINAL BOSS 1:** Full Exam 1 (50 questions) - **FINAL BOSS 2:** Full Exam 2 (50 questions) - **FINAL BOSS 3:** Full Exam 3 -- Timed (50 questions) - **Unlock at 25000 XP:** CERTIFIED ARCHITECT --- ## DAILY CHALLENGE TYPES Rotate through these to keep study sessions varied: ### Monday: FLASH ROUND Speed-drill 5 random questions from any domain. Track time. - Under 5 min: 150 XP - Under 10 min: 100 XP - Over 10 min: 50 XP ### Tuesday: DEEP DIVE Read one task statement from the exam guide thoroughly. Write a 1-paragraph summary from memory. - Accurate summary: 150 XP - Summary with errors: 75 XP ### Wednesday: BUILD DAY Hands-on exercise: build something related to the week's domain. - Working implementation: 300 XP - Partial implementation: 150 XP ### Thursday: SCENARIO SPRINT Pick one of the 6 exam scenarios. Answer 5 questions about it. - 5/5: 200 XP - 4/5: 150 XP - 3/5: 100 XP ### Friday: ANTI-PATTERN HUNT Review a code/architecture scenario and identify all anti-patterns. - All found: 200 XP - Most found: 100 XP ### Saturday: BOSS BATTLE Weekly practice test (10 questions). - See Boss Battle XP above. ### Sunday: REST or REVIEW Optional light review of weak areas. - If studied: 50 XP - Rest days do NOT break streaks --- ## ACHIEVEMENT BADGES Earn these for special accomplishments: | Badge | Requirement | Bonus XP | |-------|-------------|----------| | First Blood | Pass your first boss battle | 200 | | Perfectionist | Score 10/10 on any boss battle | 500 | | Streak Master | 14-day study streak | 300 | | Iron Will | 30-day study streak | 1000 | | Speed Demon | Complete a boss battle in under 10 min with 8+ correct | 500 | | Domain Master | Score 9+ on two consecutive tests in the same domain | 400 | | Full Clear | Pass all 10 boss battles | 1000 | | No Weakness | Score 8+ on all 10 boss battles | 2000 | | Final Boss Slayer | Pass Full Exam 1 | 1000 | | Untouchable | Score 45+ on any full exam | 3000 | | Certified | Achieve 25000 total XP | Title | --- ## THE ANTI-PATTERN BESTIARY Enemies you will face on every floor. Learn their weaknesses: ### The Prompt Whisperer (appears on floors 1, 3, 6) *"Just add it to the system prompt!"* **Weakness:** Programmatic hooks for critical business rules. Prompts are probabilistic, hooks are deterministic. ### The Sentiment Siren (appears on floor 9) *"The customer sounds angry -- escalate immediately!"* **Weakness:** Sentiment != complexity. Use explicit escalation criteria, not emotional signals. ### The Confidence Phantom (appears on floors 9, 10) *"The model says it's 95% confident, so it must be right!"* **Weakness:** Self-reported confidence is poorly calibrated. Use labeled validation sets and stratified sampling. ### The Tool Hoarder (appears on floor 4) *"Give every agent access to every tool!"* **Weakness:** 4-5 tools per agent max. Too many tools degrade selection accuracy. ### The Context Glutton (appears on floors 9, 10) *"Just dump everything into the context!"* **Weakness:** Trim verbose outputs, extract case facts, use subagents for isolation. ### The Self-Review Shadow (appears on floor 8) *"The model can review its own code objectively!"* **Weakness:** Same-session bias. Use an independent review instance. ### The Generic Error Ghost (appears on floors 4, 9) *"Operation failed. That's all you need to know."* **Weakness:** Structured error context: category, retryable flag, partial results, alternatives. ### The Aggregate Accuracy Mirage (appears on floor 10) *"97% accuracy overall -- ship it!"* **Weakness:** Check per-document-type and per-field accuracy. Aggregates mask segment failures. ### The Iteration Cap Golem (appears on floor 1) *"Set max_iterations=5 and call it a day!"* **Weakness:** Use stop_reason as primary control. Iteration caps are safety fallbacks, not primary mechanisms. ### The Loop Terminator (appears on floor 1) *"If the response says 'I'm done', stop the loop!"* **Weakness:** Never parse natural language for loop control. Check stop_reason. --- ## SCORE TRACKER Copy this to track your progress: ``` Week 1: [ ] [ ] [ ] [ ] [ ] [ ] [ ] Streak: ___ XP: ___ Week 2: [ ] [ ] [ ] [ ] [ ] [ ] [ ] Streak: ___ XP: ___ Week 3: [ ] [ ] [ ] [ ] [ ] [ ] [ ] Streak: ___ XP: ___ Week 4: [ ] [ ] [ ] [ ] [ ] [ ] [ ] Streak: ___ XP: ___ Week 5: [ ] [ ] [ ] [ ] [ ] [ ] [ ] Streak: ___ XP: ___ Week 6: [ ] [ ] [ ] [ ] [ ] [ ] [ ] Streak: ___ XP: ___ Week 7: [ ] [ ] [ ] [ ] [ ] [ ] [ ] Streak: ___ XP: ___ Week 8: [ ] [ ] [ ] [ ] [ ] [ ] [ ] Streak: ___ XP: ___ Week 9: [ ] [ ] [ ] [ ] [ ] [ ] [ ] Streak: ___ XP: ___ Week 10: [ ] [ ] [ ] [ ] [ ] [ ] [ ] Streak: ___ XP: ___ Week 11: [ ] [ ] [ ] [ ] [ ] [ ] [ ] Streak: ___ XP: ___ Week 12: [ ] [ ] [ ] [ ] [ ] [ ] [ ] Streak: ___ XP: ___ Boss Battle Scores: Test 01: __/10 | Test 02: __/10 | Test 03: __/10 Test 04: __/10 | Test 05: __/10 | Test 06: __/10 Test 07: __/10 | Test 08: __/10 | Test 09: __/10 Test 10: __/10 Full Exams: Exam 1: __/50 | Exam 2: __/50 | Exam 3: __/50 Achievements: ________________________ Total XP: ________ Current Title: ________________________ ``` --- ## HOW TO USE THIS WITH CLAUDE CODE Ask me to: - **Quiz you:** "Quiz me on Domain 1" or "Give me a flash round" - **Run a boss battle:** "Run boss battle 3" (I will present questions one at a time) - **Run a full exam:** "Run full exam 1" (I will simulate exam conditions) - **Explain a concept:** "Explain the agentic loop lifecycle" - **Grade an answer:** Explain your reasoning and I will tell you if you would have gotten credit - **Track progress:** "Update my score" -- tell me your results and I will track your XP - **Anti-pattern drill:** "Show me an anti-pattern scenario" -- I will give you a buggy architecture to diagnose Start your quest: **"Begin Week 1"**