Files
pepa-pi-bot/runtime/knowledge/schema.sql
T
mayatnikovandClaude Opus 4.7 0d97ccccfa feat(v0.3.0): paradigm shift — TimeWeb-only LLM + persistent advisor trail + improvement queue
This is the rc.4 batch the user requested:

  1. Emergency triggers (low HP + close hostile, lava-under-foot)
     bypass the long cooldown so the LLM is consulted BEFORE the bot
     dies, not after.
  2. Active manifesto need is now included in the advisor user prompt
     — the LLM picks suggestions that satisfy the bot's current
     concrete need (L2 tools_wood → "gather logs nearby" not
     "explore further").
  3. Every advisor recommendation is persisted to SQLite
     (advisor_recommendations table) with full token usage. The
     reflex marks 'applied=1' when it dispatches and updates
     outcome_ok/code when the dispatch completes. Ground truth for
     "is the LLM actually helping" lives in the DB, not in logs.
  4. Pi CLI is OUT of every background loop. coach/postmortem and
     coach/reflect now go through the same TimeWeb endpoint
     fast-advisor uses, via the shared coach/llm-call.js helper.
     Pi is reserved for manual operator commands.
  5. The LLM (postmortem, reflect, advisor) can flag "structural
     gaps" — missing skills/features the operator should implement.
     These land in the new improvement_requests table. Dedup by
     title bumps `votes` instead of inserting duplicates so the
     queue doesn't bloat. Operator views via
     `node scripts/list-improvements.js`.
  6. A deterministic trigger-tuner runs hourly: reads 24h of
     recommendation stats, flags triggers whose success rate is
     below 25% (sample ≥ 5) or whose prompts are expensive (>1000
     input tokens) with mediocre payoff. Improvements get
     source="tuner", category="tuning". No LLM call.

New files:
  runtime/coach/llm-call.js        — askAnalytical() helper
  runtime/coach/trigger-tuner.js   — stats → improvements
  runtime/coach/trigger-tuner.test.js
  scripts/list-improvements.js     — operator CLI

Schema additions:
  advisor_recommendations: id, ts, trigger_reason, planned_skill,
    recommended_skill, action, rationale, active_need, tokens_in,
    tokens_out, latency_ms, applied, outcome_ok, outcome_code, outcome_at
  improvement_requests: id, ts, source, category, title, description,
    context, priority, status, duplicate_of, votes, implemented_at, notes

Renamed env-var consumers:
  Pi-coach drainOnce({ askPi })   → drainOnce({ askAnalyticalFn? })
  Pi-reflect runOnce({ askPi })   → runOnce({ askAnalyticalFn? })
  bot.js attachCoach/attachReflect no longer pass askPi
  attachTuner() added to bot.js spawn handler
  lessons.source 'pi-coach'   → 'timeweb-coach'
  lessons.source 'pi-reflect' → 'timeweb-reflect'

Token cost measured live:
  ~705 input + 45 output = ~750 total per advisor call
  worst case @ 6 calls/hour rate cap = ~108K tokens/day
  OpenAI gpt-5-mini reference price: ~$0.60/month

Operator usage:
  node scripts/list-improvements.js                # open queue
  node scripts/list-improvements.js --stats        # advisor performance
  node scripts/list-improvements.js --done 17 "shipped in 0.3.1"
  node scripts/list-improvements.js --reject 18 "duplicate"

Tests: 360 green (was 332, +28 new).

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-27 19:12:48 +03:00

242 lines
11 KiB
SQL

-- pepa-pi-bot knowledge schema v1
-- Single-process SQLite store at state/<host>/knowledge.db.
-- Idempotent: applied on every boot. Migrations go below the CREATE TABLE
-- block, gated by schema_version.
CREATE TABLE IF NOT EXISTS schema_version (
version INTEGER NOT NULL PRIMARY KEY,
applied_at INTEGER NOT NULL
);
----------------------------------------------------------------------
-- Recipes (seeded from docs/minecraft-recipes.json + augmented by wiki)
----------------------------------------------------------------------
CREATE TABLE IF NOT EXISTS recipes (
id INTEGER PRIMARY KEY AUTOINCREMENT,
name TEXT NOT NULL UNIQUE,
shape TEXT NOT NULL, -- JSON array of rows
shapeless INTEGER NOT NULL DEFAULT 0,
yields INTEGER NOT NULL DEFAULT 1,
requires_table INTEGER NOT NULL DEFAULT 1, -- 0=hand,1=table,2=furnace,3=smithing
source TEXT,
source_url TEXT,
updated_at INTEGER NOT NULL
);
CREATE INDEX IF NOT EXISTS idx_recipes_name ON recipes(name);
----------------------------------------------------------------------
-- Mob intel — what to do when you see a mob
----------------------------------------------------------------------
CREATE TABLE IF NOT EXISTS mob_intel (
id INTEGER PRIMARY KEY AUTOINCREMENT,
name TEXT NOT NULL UNIQUE,
hostility TEXT NOT NULL, -- 'hostile' | 'neutral' | 'passive' | 'tamable'
threat_level INTEGER NOT NULL, -- 1..5
approach_range REAL, -- blocks at which it engages
burns_in_sun INTEGER NOT NULL DEFAULT 0,
ranged INTEGER NOT NULL DEFAULT 0,
weakness TEXT,
drops TEXT, -- JSON array of names
verdict_no_weapon TEXT, -- 'flee' | 'shelter' | 'avoid' | 'pillar'
verdict_with_sword TEXT, -- 'kite' | 'attack' | 'avoid'
notes TEXT,
source TEXT,
updated_at INTEGER NOT NULL
);
CREATE INDEX IF NOT EXISTS idx_mob_name ON mob_intel(name);
----------------------------------------------------------------------
-- Block intel — what tool, what drops, lighting
----------------------------------------------------------------------
CREATE TABLE IF NOT EXISTS block_intel (
id INTEGER PRIMARY KEY AUTOINCREMENT,
name TEXT NOT NULL UNIQUE,
required_tool TEXT, -- 'any'|'wood_pickaxe'|'stone_pickaxe'|'iron_pickaxe'|'shovel'|'axe'
drops TEXT, -- JSON array
light_emit INTEGER DEFAULT 0,
walkable INTEGER DEFAULT 1,
notes TEXT,
source TEXT,
updated_at INTEGER NOT NULL
);
CREATE INDEX IF NOT EXISTS idx_block_name ON block_intel(name);
----------------------------------------------------------------------
-- Lessons — generalised "what to do / what to avoid" learned over time
----------------------------------------------------------------------
CREATE TABLE IF NOT EXISTS lessons (
id INTEGER PRIMARY KEY AUTOINCREMENT,
ts INTEGER NOT NULL,
text TEXT NOT NULL,
category TEXT NOT NULL, -- 'combat'|'pathing'|'crafting'|'survival'|'social'|'self-improve'
trigger_skill TEXT,
trigger_hostile TEXT,
trigger_situation TEXT, -- coarse hash key from scenario-memory
avoid_skill TEXT,
prefer_skill TEXT,
confidence REAL NOT NULL DEFAULT 0.5,
applied_count INTEGER NOT NULL DEFAULT 0,
succeeded_count INTEGER NOT NULL DEFAULT 0,
source TEXT NOT NULL, -- 'postmortem'|'pi-coach'|'wiki'|'operator'|'rule'
source_ref TEXT
);
CREATE INDEX IF NOT EXISTS idx_lessons_category ON lessons(category);
CREATE INDEX IF NOT EXISTS idx_lessons_skill ON lessons(trigger_skill);
CREATE INDEX IF NOT EXISTS idx_lessons_hostile ON lessons(trigger_hostile);
CREATE INDEX IF NOT EXISTS idx_lessons_ts ON lessons(ts);
----------------------------------------------------------------------
-- Death events — captured by coach/postmortem.js on every death
----------------------------------------------------------------------
CREATE TABLE IF NOT EXISTS deaths (
id INTEGER PRIMARY KEY AUTOINCREMENT,
ts INTEGER NOT NULL,
x REAL, y REAL, z REAL,
cause TEXT, -- 'hostile'|'fall'|'lava'|'drowning'|'starvation'|'suffocation'|'other'|'unknown'
hostile TEXT,
last_skill TEXT,
last_skill_code TEXT,
hp_at_death REAL,
food_at_death REAL,
inventory_lost TEXT, -- JSON
context_blob TEXT, -- JSON: last 30s of events
analysed INTEGER NOT NULL DEFAULT 0
);
CREATE INDEX IF NOT EXISTS idx_deaths_ts ON deaths(ts);
CREATE INDEX IF NOT EXISTS idx_deaths_analysed ON deaths(analysed);
CREATE INDEX IF NOT EXISTS idx_deaths_cause ON deaths(cause);
----------------------------------------------------------------------
-- Pi-extracted post-mortems linking back to deaths
----------------------------------------------------------------------
CREATE TABLE IF NOT EXISTS postmortems (
id INTEGER PRIMARY KEY AUTOINCREMENT,
death_id INTEGER NOT NULL,
ts INTEGER NOT NULL,
cause TEXT,
lesson TEXT,
next_action TEXT,
raw_response TEXT,
source TEXT, -- 'pi' | 'rule'
FOREIGN KEY (death_id) REFERENCES deaths(id)
);
CREATE INDEX IF NOT EXISTS idx_postmortems_death ON postmortems(death_id);
----------------------------------------------------------------------
-- Points of interest — queryable spatial memory
----------------------------------------------------------------------
CREATE TABLE IF NOT EXISTS poi (
id INTEGER PRIMARY KEY AUTOINCREMENT,
kind TEXT NOT NULL, -- 'tree'|'ore'|'water'|'mob_spawner'|'danger'|'foreign_build'|'base'|'chest'
name TEXT,
x REAL NOT NULL, y REAL NOT NULL, z REAL NOT NULL,
cell_x INTEGER NOT NULL,
cell_z INTEGER NOT NULL,
ts INTEGER NOT NULL,
expires_at INTEGER,
notes TEXT
);
CREATE INDEX IF NOT EXISTS idx_poi_cell ON poi(cell_x, cell_z);
CREATE INDEX IF NOT EXISTS idx_poi_kind ON poi(kind);
----------------------------------------------------------------------
-- Cached wiki pages (rc.2)
----------------------------------------------------------------------
CREATE TABLE IF NOT EXISTS wiki_pages (
id INTEGER PRIMARY KEY AUTOINCREMENT,
slug TEXT NOT NULL UNIQUE,
url TEXT NOT NULL,
body TEXT,
etag TEXT,
fetched_at INTEGER NOT NULL,
expires_at INTEGER NOT NULL
);
----------------------------------------------------------------------
-- Full chat log (durable; per-speaker LRU in memory is unchanged)
----------------------------------------------------------------------
CREATE TABLE IF NOT EXISTS chat_log (
id INTEGER PRIMARY KEY AUTOINCREMENT,
ts INTEGER NOT NULL,
direction TEXT NOT NULL, -- 'in' | 'out'
speaker TEXT,
text TEXT NOT NULL,
intent TEXT,
replied_with TEXT
);
CREATE INDEX IF NOT EXISTS idx_chat_speaker ON chat_log(speaker);
CREATE INDEX IF NOT EXISTS idx_chat_ts ON chat_log(ts);
----------------------------------------------------------------------
-- Self-rewrite audit
----------------------------------------------------------------------
CREATE TABLE IF NOT EXISTS code_changes (
id INTEGER PRIMARY KEY AUTOINCREMENT,
ts INTEGER NOT NULL,
proposal_slug TEXT,
files TEXT, -- JSON array
diff_hash TEXT,
outcome TEXT, -- 'applied'|'rolled_back'|'rejected'
notes TEXT
);
----------------------------------------------------------------------
-- Advisor recommendations (v0.3.0+ fast LLM trail)
-- Every time runtime/coach/advisor-trigger.js asks the fast LLM and
-- the answer is cached on ctx, we write a row here. When the reflex
-- consumes the recommendation and dispatches, we attach the dispatch
-- result later via outcome_ok / outcome_code. The history is the
-- ground truth for trigger-tuner.js stats and for the operator's
-- "what is the LLM suggesting and is it actually helping" question.
----------------------------------------------------------------------
CREATE TABLE IF NOT EXISTS advisor_recommendations (
id INTEGER PRIMARY KEY AUTOINCREMENT,
ts INTEGER NOT NULL,
trigger_reason TEXT NOT NULL, -- 'wedged_*', 'repeat_*', 'preempt_retry_*', 'emergency_*'
planned_skill TEXT, -- what manifesto/curriculum was about to dispatch
recommended_skill TEXT, -- what the LLM said to do instead
action TEXT NOT NULL, -- 'switch_skill' | 'continue' | 'wait'
rationale TEXT,
active_need TEXT, -- 'L2 tools_wood' etc.
tokens_in INTEGER,
tokens_out INTEGER,
latency_ms INTEGER,
applied INTEGER NOT NULL DEFAULT 0, -- 1 if reflex actually dispatched recommended_skill
outcome_ok INTEGER, -- NULL until dispatch finishes
outcome_code TEXT,
outcome_at INTEGER
);
CREATE INDEX IF NOT EXISTS idx_advisor_ts ON advisor_recommendations(ts);
CREATE INDEX IF NOT EXISTS idx_advisor_trigger ON advisor_recommendations(trigger_reason);
CREATE INDEX IF NOT EXISTS idx_advisor_outcome ON advisor_recommendations(outcome_ok);
----------------------------------------------------------------------
-- Improvement requests (v0.3.0+)
-- The LLM (postmortem / reflect / advisor) can flag situations where
-- the bot lacked the right skill or feature. Instead of trying to
-- self-patch (which we explicitly disabled), it writes an entry here.
-- The operator reads `scripts/list-improvements.js` and decides what
-- to implement. Implemented entries get marked so the bot stops
-- re-flagging the same gap.
----------------------------------------------------------------------
CREATE TABLE IF NOT EXISTS improvement_requests (
id INTEGER PRIMARY KEY AUTOINCREMENT,
ts INTEGER NOT NULL,
source TEXT NOT NULL, -- 'postmortem'|'reflect'|'advisor'|'tuner'|'manual'
category TEXT, -- 'skill'|'tuning'|'perception'|'planning'|'social'|'other'
title TEXT NOT NULL,
description TEXT,
context TEXT, -- JSON: position, snapshot tail, related lesson ids
priority INTEGER NOT NULL DEFAULT 3, -- 1..5 (1=urgent, 5=nice-to-have)
status TEXT NOT NULL DEFAULT 'open', -- 'open'|'in_progress'|'implemented'|'rejected'|'duplicate'
duplicate_of INTEGER, -- another row id if dup
votes INTEGER NOT NULL DEFAULT 1, -- bumped each time the bot re-flags same gap
implemented_at INTEGER,
notes TEXT,
FOREIGN KEY (duplicate_of) REFERENCES improvement_requests(id)
);
CREATE INDEX IF NOT EXISTS idx_improvements_status ON improvement_requests(status);
CREATE INDEX IF NOT EXISTS idx_improvements_priority ON improvement_requests(priority);
CREATE INDEX IF NOT EXISTS idx_improvements_source ON improvement_requests(source);
CREATE INDEX IF NOT EXISTS idx_improvements_ts ON improvement_requests(ts);