miti99bot

tiennm99/miti99bot

Fork 0

mirror of https://github.com/tiennm99/miti99bot.git synced 2026-04-28 10:20:35 +00:00

Commit Graph

Author	SHA1	Message	Date
tiennm99	3be799d68a	chore: project cleanup — purge stale function-calling refs + sync docs Followed code-reviewer audit. Findings applied: - twentyq/README.md, twentyq/index.js header — claimed "function calling" + ANSWER_FUNCTION_SCHEMA / submit_answer; rewrite to JSON-in-content matching what the code actually does. Added generateRoundStart line. - wrangler.toml [ai] comment — list both bge-m3 (semantle/doantu) AND gemma-4 (twentyq) consumers; drop neuron math that no longer matched. - scripts/stub-kv.js — drop reference to nonexistent REGISTER_DRYRUN flag. - twentyq/ai-client.redactSecret — strip dead "if (out.length > 0)" branch (String.replace cannot produce empty string from the inputs we pass). - handlers.test.js — drop noise saveGame() before "no games" stats assert; add ai.run call-count guards on two-AI-call flows. - docs/codebase-summary.md — full rewrite of Active Modules table (semantle/doantu/lolschedule/twentyq were missing); fix vitest 2→4 + wrangler 3→4 versions; replace stale 200-test count with current ~450. - docs/architecture.md — file tree includes lolschedule/semantle/doantu/ twentyq + cron-dispatcher + sql-store* + scripts/migrate.js; moduleRegistry snippet matches src/modules/index.js. - docs/todo.md — entire file obsolete (D1 UUID populated, cron live). Deleted. Tests: 449 pass, lint clean.	2026-04-24 18:30:25 +07:00
tiennm99	f6ab94ffb0	feat(twentyq): LLM-generated category + initial hint from bare keyword seeds Previously seeds carried hand-curated {category, target, initialHint}. Now SEEDS is a flat string[] of keywords — at round-start, the model generates {category, initialHint} on the fly. Benefits: - adding a seed is trivial (just append a word) - every round gets a fresh cryptic opener (varies across plays of the same word) - HINT STYLE rules apply to the opening hint too, so the initial clue isn't a definitional giveaway Implementation: - prompts.buildStartRoundPrompt(target) — with good/bad examples - ai-client.generateRoundStart(env, target) — same JSON-in-content approach as judge(), with defensive fallbacks + redactSecret - handlers.startFreshGame now async; surfaces roundstart errors via the existing UPSTREAM_FAIL path Tests: 449 pass (5 new for generateRoundStart, 1 for roundstart error path).	2026-04-24 16:26:55 +07:00
tiennm99	5b12650906	feat(twentyq): add reverse-Akinator yes/no game module powered by Workers AI - seeded 54 objects across 6 categories (instrument, animal, food, vehicle, sport, household) - @cf/google/gemma-4-26b-a4b-it judges via function calling; returns {is_guess, answer, hint} - pre-AI validator rejects open-ended questions; handler dedups exact repeats - secret-redacting hint filter as defense-in-depth - 86 new vitest tests (seeds, state, validator, ai-client, handlers, render)	2026-04-24 14:37:23 +07:00

Author

SHA1

Message

Date

tiennm99

3be799d68a

chore: project cleanup — purge stale function-calling refs + sync docs

Followed code-reviewer audit. Findings applied:
- twentyq/README.md, twentyq/index.js header — claimed "function calling"
  + ANSWER_FUNCTION_SCHEMA / submit_answer; rewrite to JSON-in-content
  matching what the code actually does. Added generateRoundStart line.
- wrangler.toml [ai] comment — list both bge-m3 (semantle/doantu) AND
  gemma-4 (twentyq) consumers; drop neuron math that no longer matched.
- scripts/stub-kv.js — drop reference to nonexistent REGISTER_DRYRUN flag.
- twentyq/ai-client.redactSecret — strip dead "if (out.length > 0)" branch
  (String.replace cannot produce empty string from the inputs we pass).
- handlers.test.js — drop noise saveGame() before "no games" stats assert;
  add ai.run call-count guards on two-AI-call flows.
- docs/codebase-summary.md — full rewrite of Active Modules table
  (semantle/doantu/lolschedule/twentyq were missing); fix vitest 2→4 +
  wrangler 3→4 versions; replace stale 200-test count with current ~450.
- docs/architecture.md — file tree includes lolschedule/semantle/doantu/
  twentyq + cron-dispatcher + sql-store* + scripts/migrate.js;
  moduleRegistry snippet matches src/modules/index.js.
- docs/todo.md — entire file obsolete (D1 UUID populated, cron live).
  Deleted.

Tests: 449 pass, lint clean.

2026-04-24 18:30:25 +07:00

tiennm99

f6ab94ffb0

feat(twentyq): LLM-generated category + initial hint from bare keyword seeds

Previously seeds carried hand-curated {category, target, initialHint}.
Now SEEDS is a flat string[] of keywords — at round-start, the model
generates {category, initialHint} on the fly. Benefits:
- adding a seed is trivial (just append a word)
- every round gets a fresh cryptic opener (varies across plays of the
  same word)
- HINT STYLE rules apply to the opening hint too, so the initial clue
  isn't a definitional giveaway

Implementation:
- prompts.buildStartRoundPrompt(target) — with good/bad examples
- ai-client.generateRoundStart(env, target) — same JSON-in-content
  approach as judge(), with defensive fallbacks + redactSecret
- handlers.startFreshGame now async; surfaces roundstart errors via the
  existing UPSTREAM_FAIL path

Tests: 449 pass (5 new for generateRoundStart, 1 for roundstart error path).

2026-04-24 16:26:55 +07:00

tiennm99

5b12650906

feat(twentyq): add reverse-Akinator yes/no game module powered by Workers AI

- seeded 54 objects across 6 categories (instrument, animal, food, vehicle, sport, household)
- @cf/google/gemma-4-26b-a4b-it judges via function calling; returns {is_guess, answer, hint}
- pre-AI validator rejects open-ended questions; handler dedups exact repeats
- secret-redacting hint filter as defense-in-depth
- 86 new vitest tests (seeds, state, validator, ai-client, handlers, render)

2026-04-24 14:37:23 +07:00

3 Commits