Known answers for the bot: the guide's encounter answers and the bot's own results, by choice text #39

Merged
kathleen merged 1 commits from atlas/known-answers into main 2026-11-19 14:00:18 +00:00
Owner

Part of #35 (bot side: DragonKind/DragonKind#4).

The bot asks this for screens that neither the policy nor the PDF guide covers.

  • analysis/known/build_known.py builds data/public/known_answers.json from pdf_encounters.json and tree.json, keyed by normalised choice text (the same normalize() as the bot transform).
    • Guide entries hold the encounter, its colours, the end, on_path, and whether the answer is unconfirmed.
    • Bot entries hold n, died and bonded.
    • --check is in test:py; the builder is in npm run policy and in rebuild, after the screen tree.
  • core/solver.js knownAnswer(known, offered, {target, sceneColour}) returns a pick when the evidence is clear (strong / weak) plus a verdict for every option (good / bad / unknown), which the bot hands to Clef.
    • "Bad" means the guide shows incineration, or the bot died on that choice at least 75% of the time over 2 or more runs.
    • The guide's paraphrased labels are matched by content-word containment.
  • push-policy.yml vendors known_answers.json to the bot and triggers when it changes.
  • Tests: tests/core/known.test.js (4).

No game prose: only short choice labels and counts.

🤖 Generated with Claude Code

Part of #35 (bot side: DragonKind/DragonKind#4). The bot asks this for screens that neither the policy nor the PDF guide covers. - **`analysis/known/build_known.py`** builds `data/public/known_answers.json` from `pdf_encounters.json` and `tree.json`, keyed by normalised choice text (the same `normalize()` as the bot transform). - Guide entries hold the encounter, its colours, the end, `on_path`, and whether the answer is unconfirmed. - Bot entries hold `n`, `died` and `bonded`. - `--check` is in `test:py`; the builder is in `npm run policy` and in `rebuild`, after the screen tree. - **`core/solver.js` `knownAnswer(known, offered, {target, sceneColour})`** returns a pick when the evidence is clear (`strong` / `weak`) plus a verdict for every option (good / bad / unknown), which the bot hands to Clef. - "Bad" means the guide shows incineration, or the bot died on that choice at least 75% of the time over 2 or more runs. - The guide's paraphrased labels are matched by content-word containment. - **`push-policy.yml`** vendors `known_answers.json` to the bot and triggers when it changes. - Tests: `tests/core/known.test.js` (4). No game prose: only short choice labels and counts. 🤖 Generated with [Claude Code](https://claude.com/claude-code)
kathleen added 1 commit 2026-11-19 13:58:09 +00:00
analysis/known/build_known.py writes data/public/known_answers.json from pdf_encounters.json and tree.json;
solver.js knownAnswer(known, offered, { target, sceneColour }) matches a screen's choices against it (the guide's
paraphrases by their words, the bot's texts exactly) and returns a pick when one choice is clearly on the guide's
recommended path to the target, plus the evidence for every choice to hand to Clef. Rebuilt daily, checked in CI,
and sent to the bot with the route model.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UCzTd31GphpMXDhQqWrCMQ
kathleen merged commit e06fd46e15 into main 2026-11-19 14:00:18 +00:00
kathleen deleted branch atlas/known-answers 2026-11-19 14:00:18 +00:00
Sign in to join this conversation.
No Reviewers
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: DragonKind/threshing-atlas#39