Why did sign-ups drop?
Quest: Find out why sign-ups fell last week using PostHog and a local data export
Equipment

- Quest
- Find out why sign-ups fell last week using PostHog and a local data export
- Hero
- Claude CodeThe agent that runs the quest
- Weapon
Claude Sonnet 5.5Longsword: Balanced speed and depthMana rate 10 per 1k tokens Mana per run without spells: 1,049 mana (293 runs, 22 quests).- Charge
- MediumReasoning effort
- Author
- The curator
PostHog shows the funnel and recent errors, DuckDB loads the exported file and queries it locally, and the Data plugin charts the answer and runs its validation pass before you share it.
Can't be appraised: Proving Grounds can't equip orbs (MCP servers).
Skill tree
ScoutGather context
CraftMake the thing
TestCheck the work
Nothing equipped here
ShipDeliver it
Nothing equipped here
Quest replay
Quest: "Find out why sign-ups fell last week using PostHog and a local data export". Four waves, then the Backlog Dragon.
Foes
- Fog WraithHaze Poison Ward Insight Flicker Pinned Enraged
- Riddle SphinxHaze Poison Ward Insight Flicker Pinned Enraged
- Bug SlimeHaze Poison Ward Insight Flicker Pinned Enraged
- Flaky WispHaze Poison Ward Insight Flicker Pinned Enraged
- Bug SlimeHaze Poison Ward Insight Flicker Pinned Enraged
- Gate WardenHaze Poison Ward Insight Flicker Pinned Enraged
- Backlog DragonHaze Poison Ward Insight Flicker Pinned Enraged
- Bug SlimeHaze Poison Ward Insight Flicker Pinned Enraged
Party · Context
- Claude CodeAlchemist100/100Haze Poison Ward Insight Flicker Pinned Enraged
- ExploreSubagent0/40Haze Poison Ward Insight Flicker Pinned Enraged
The Fog Wraith steps up.
What really happensGather context
Claude Code: Context 100/100.
Claude Code sends Explore, a fox scout, into the fog.
What really happensClaude Code can hand a codebase search to its built-in Explore subagent, which works in its own context window.
Explore: Context 40/40.
Explore sniffs out the right files.
What really happensThe subagent reads and searches. None of that output lands in the main context.
Fog Wraith takes 21.
A watchtower that answers product questions from PostHog.
What really happensThe agent queries insights in plain words, such as sign-ups this week or the checkout funnel conversion rate.
Fog Wraith takes 38.
The Fog Wraith breathes haze. Swings start to miss.
What really happensWithout current docs, the agent guesses at APIs that may have changed.
Explore takes 3. Explore: Context 37/40. Explore: Haze.
Explore sniffs out the right files.
Fog Wraith takes 0.
A pocket duck engine that runs SQL on your local data.
What really happenslist_databases, list_tables and list_columns show what is attached before any query runs.
Fog Wraith takes 1. Fog Wraith falls.
Fresh pages burn the haze away.
What really happensCurrent docs replace guesses with facts.
Explore returns with a short report.
What really happensOnly the subagent's final summary comes back to the main agent.
Claude Code takes 2. Claude Code: Context 98/100. Claude Code: Insight. Explore returns.
The Riddle Sphinx steps up.
What really happensMake the thing
A satchel of analyst scrolls for questions your tables can answer.
What really happensexplore-data profiles a table: row counts, nulls, cardinality and odd values worth a closer look.
Riddle Sphinx takes 55 (critical hit).
The sphinx asks a riddle. A wrong guess costs dearly.
What really happensAn answer guessed from a hunch instead of a query sends the work the wrong way.
Claude Code takes 6. Claude Code: Context 92/100.
Claude Code swings the Longsword.
Riddle Sphinx takes 45 (critical hit). Riddle Sphinx falls.
The Bug Slime and the Flaky Wisp step up.
What really happensCheck the work
Flaky Wisp: Flicker.
The wisp flickers out of reach.
What really happensA flaky test passes on one run and fails on the next.
Flaky Wisp avoids the hit.
The Flaky Wisp uses Flicker.
Claude Code takes 3. Claude Code: Context 89/100.
The Bug Slime uses Regress.
Claude Code takes 3. Claude Code: Context 86/100.
Claude Code swings the Longsword.
Flaky Wisp takes 40. Flaky Wisp falls.
The Bug Slime uses Regress.
Claude Code takes 3. Claude Code: Context 83/100.
Claude Code swings the Longsword.
Bug Slime takes 40.
The slime splits. A squashed bug crawls back.
What really happensA fix in one place breaks something that used to work.
Claude Code swings the Longsword.
Bug Slime takes 5. Bug Slime falls.
The Bug Slime uses Regress.
Claude Code takes 3. Claude Code: Context 80/100.
Claude Code swings the Longsword.
Bug Slime takes 18. Bug Slime falls.
The Gate Warden steps up.
What really happensDeliver it
Gate Warden: Barrier.
The Backlog Dragon guards the quest: "Find out why sign-ups fell last week using PostHog and a local data export".
What really happensIt stands for the whole quest, every task on the list.
Claude Code swings the Longsword.
Gate Warden is behind a barrier.
The dragon breathes scope creep across the party.
What really happensNew asks pile onto a task while it is underway.
Claude Code takes 5. Claude Code: Context 75/100.
The warden raises the gate. Nothing passes until the checks are green.
What really happensA release waits until tests, builds and reviews pass.
Claude Code swings the Longsword.
Gate Warden is behind a barrier.
The Backlog Dragon uses Claw.
Claude Code takes 6. Claude Code: Context 69/100.
Claude Code swings the Longsword.
Gate Warden takes 40.
The checks turn green and the gate opens.
What really happensTests, build and review pass, so the release can go.
The Gate Warden swings the gate shut.
Claude Code takes 5. Claude Code: Context 64/100.
The Backlog Dragon uses Scope creep.
Claude Code takes 5. Claude Code: Context 59/100.
Claude Code swings the Longsword.
Gate Warden takes 20. Gate Warden falls.
Claude Code swings the Longsword.
Backlog Dragon takes 36.
The Backlog Dragon uses Claw.
Claude Code takes 6. Claude Code: Context 53/100.
Claude Code swings the Longsword.
Backlog Dragon takes 41.
The Backlog Dragon uses Scope creep.
Claude Code takes 5. Claude Code: Context 48/100.
Claude Code swings the Longsword.
Backlog Dragon takes 40.
The dragon roars, and another task crawls out of the backlog.
What really happensFinishing one task often surfaces the next one on the list.
Claude Code swings the Longsword.
Bug Slime takes 25. Bug Slime falls.
The Backlog Dragon uses Scope creep.
Claude Code takes 5. Claude Code: Context 43/100.
Claude Code swings the Longsword.
Backlog Dragon takes 38.
Wounded, the dragon rages.
What really happensThe last stretch of a big job is often the hardest.
Backlog Dragon: Enraged.
Claude Code swings the Longsword.
Backlog Dragon takes 42.
The Backlog Dragon uses Claw.
Claude Code takes 8. Claude Code: Context 35/100.
Claude Code swings the Longsword.
Backlog Dragon takes 23. Backlog Dragon falls.
The Backlog Dragon falls. Quest complete.
What really happensEvery equipped spell did its part of "Find out why sign-ups fell last week using PostHog and a local data export".
- Claude Code took 20 turns to clear four waves.
- Context ended at 35 of 100.
- No Ship spell, so that wave ran on basic attacks.
- No Test spell, so a regression slipped back in.
- Summons took 3 damage in their own context and brought back 1 report.
Copy this loadout
Prompt for Claude Code
Paste this into Claude Code. It installs each spell, leaves API keys and sign-ins to you, and ends with the command that restarts it with this weapon and charge.
Set up these tools for me in Claude Code.
Loadout: https://vibe-rpg-simulator.vercel.app/loadouts/signup-drop-analysis/
1. PostHog
Source: https://github.com/PostHog/ai-plugin
Run:
claude plugin install posthog
2. DuckDB / MotherDuck MCP
Source: https://github.com/motherduckdb/mcp-server-motherduck
Run:
claude mcp add --scope user duckdb --transport stdio -- uvx mcp-server-motherduck --db-path :memory: --read-write --allow-switch-databases
3. Anthropic Data
Source: https://github.com/anthropics/knowledge-work-plugins/tree/main/data
Run:
claude plugin marketplace add anthropics/knowledge-work-plugins
claude plugin install data@knowledge-work-plugins
How to do it:
- Skip anything that is already installed. Run the rest in order and check that each step worked.
- Never ask me to paste an API key or token into this chat. When a command needs one, such as YOUR_API_KEY, give me the full command to run in my own terminal.
- Ask me for any other value you can't work out, such as a project path.
- If a command asks a question or needs a browser sign-in or an open app, tell me what to do and wait.
- If a command fails, follow the install steps in the tool's source for Claude Code instead. Show me any command that is not on this list before you run it, and install nothing else.
When you're done, tell me what is ready and what still needs me. New tools load in a new session, so finish by telling me to restart Claude Code with Claude Sonnet 5.5 and reasoning effort medium:
claude --model sonnet --effort medium
Terminal commands
Or paste these lines into a terminal. The first part installs every spell for Claude Code; the last line starts it with the same weapon and charge. Lines starting with # are notes or commands you type inside the agent.
# Loadout: Why did sign-ups drop?
# Quest: Find out why sign-ups fell last week using PostHog and a local data export
# Hero: Claude Code · Weapon: Claude Sonnet 5.5 · Charge: Medium
# From https://vibe-rpg-simulator.vercel.app/loadouts/signup-drop-analysis/
# 1. Learn the spells
# Scout: PostHog
claude plugin install posthog
# Scout: DuckDB / MotherDuck MCP
claude mcp add --scope user duckdb --transport stdio -- uvx mcp-server-motherduck --db-path :memory: --read-write --allow-switch-databases
# Craft: Anthropic Data
claude plugin marketplace add anthropics/knowledge-work-plugins
claude plugin install data@knowledge-work-plugins
# 2. Start the hero with this weapon and charge
claude --model sonnet --effort medium








