written 2026-09-26 12:45:36 · claude-opus-5
Your directive: You are a fork with the whole design conversation in context. The developer has answered many open questions. The documentation validator now reports issue files that still cite questions she has answered, and those issues need their citations and their bodies brought up to date.
Project root: /mnt/cmdo/ritz/games/tq/ai-stuff/double-diaper-dungeon.
FIRST read /tmp/claude-1000/-mnt-cmdo-ritz-games-tq-ai-stuff-double-diaper-dungeon/f76814ba-6c5e-46ea-90d1-128d98043fd5/scratchpad/her-answers-digest.md — that is her settled design, in her own words with the reading beside each. It is the authority. Also read docs/001-what-this-game-is.md for the fixed vocabulary.
YOU OWN EXACTLY THESE TWELVE FILES. Edit only these:
Do NOT touch any other file, any document under docs/, the balance table, or docs/020-open-questions.md. Do not run git. Do not rename any file.
FOR EACH ISSUE, two jobs.
One: the Open questions row. Remove every question id she has now answered. Keep the ones still open. If none remain, write "none". The ids she answered that touch your files are C1, C2, C3, C4, C9, C11, E1, E2, E7, E9, E12. Anything else stays.
Two: the body. Where her answer changes what the issue is meant to build, rewrite the Intended behavior and the Suggested implementation steps so the blueprint is correct. Do not add a worklog, do not add a "changed on this date" note; the issue is a blueprint and it simply says the right thing now. Keep the format: the header table, Current behavior, Intended behavior, Suggested implementation steps, Related documents and tools, Still open.
The ones whose bodies genuinely change:
When done, reply with: each file and its word count, the balance-table entries you named that do not exist yet (dotted name and one line of meaning), and any new open questions (topic plus one sentence). Nothing else.
I'll update the twelve issue files to match her settled answers — citations and bodies both.
Now I'll rewrite them. Starting with 106, which her answer replaces outright.
All twelve pass the validator with no stale citations.
Files and word counts
| File | Words |
|---|---|
| issues/104-the-hex-map-is-built-by-a-tool.md | 1483 |
| issues/106-the-ghost-room-is-the-farthest-by-an-unlock-aware-flood.md | 1670 |
| issues/107-a-room-fills-with-monsters-when-nobody-is-in-it.md | 1586 |
| issues/410-reinforcements-from-connected-rooms.md | 923 |
| issues/501-the-priority-table.md | 1464 |
| issues/504-exploring-and-heading-for-the-ghost-room.md | 1491 |
| issues/505-the-wait-the-call-and-the-safety-estimate.md | 1347 |
| issues/506-toilets-seeking-sensation-and-the-retry-lock.md | 1117 |
| issues/508-going-home.md | 912 |
| issues/713-levels-up-and-down.md | 1781 |
| issues/714-the-intro.md | 967 |
| issues/phase-1-progress.md | 645 |
Balance entries named that do not exist yet
map.locked_door_detour_multiplier — what a key detour costs per room of distance when pricing a locked door; two, being out and back.map.levels_fewer_than_ghosts — how much smaller the reachable span of level indices is than the ghost catalogue's length.map.difficulty_per_step_from_ground — how much harder a level gets per step of absolute index.map.reward_per_step_from_ground — how much better the rewards get per step; deliberately a separate curve from the one above.room.infestation_interval — ticks between one room becoming infested. Supersedes room.infestation_ticks, which described the per-room countdown that no longer exists.room.infestation_choice — the word selecting how the drip picks its candidate; ruled as weighting by time sat empty.basecamp.build_duration — ticks the party spends building the camp beside the ghost room.basecamp.build_cost — gold the build takes from the guild's purse.explore.never_visited_weight — the weight a never-entered neighbour carries in the wander.explore.weight_per_tick_unvisited — how fast a visited neighbour's weight grows with time since anyone was there.monster.tier_worth — how many monsters of the tier below one monster of a tier counts as, in the safety estimate; her figure is two.New open questions