feat(runtime): persistent memory — world-journal + scenario-memory
Closes a structural gap: the bot now actually REMEMBERS what it
discovered and what it tried. Two stores live under state/<host>/ and
are wired in automatically.
runtime/world-journal.js
- Append-only JSONL of discovered points (chopped, placed, base,
shelter, farm, dead_end). Indexed by 16-block spatial grid; O(neighbors)
nearest() lookups; 6 h age prune; 10k line ceiling with trim.
- leanestQuadrant({x,z}) reports the quadrant the bot has the FEWEST
markers in — used by explore.far to circle rather than retread.
- summary() exposed for the stuck-incident proposal body.
runtime/scenario-memory.js
- Sliding window of (skillId, situationHash, code, ok, detail) tuples.
- situationHash() is a coarse fingerprint (16x8x16 cell + day/night +
food/hp bucket + inv key set + closest hostile). So "same kind of
place + same kind of state" matches.
- shouldSkip({skillId, situation}) → true after ≥3 failures within 30
min UNLESS a more-recent success in the same situation un-locks it.
- recentTailFor() exposed for the stuck-incident body.
Wiring (runtime/bot.js):
- dispatchAction captures situationHash BEFORE the action runs and
records (skillId, situation, code, ok) after — failures are attributed
to the dispatch-time state, not the partial-effect state.
- worldDelta fields (choppedAt, minedAt, placedAt, baseAt, shelterAt,
plantedAt, harvestedAt, tilledAt) auto-flow into the journal.
- no_target + silent_dig_failure also write dead_end markers.
Scheduler / skills now consume memory:
- reflex.js curriculum reflex calls memory.shouldSkip — if the same
(skill, situation) failed 3+ times recently, auto-converts to a
wander hint so the bot leaves and tries elsewhere.
- explore.far calls journal.leanestQuadrant when multiple cardinal
directions are walkable and prefers the less-explored one.
- gather.logs walks to the nearest known "chopped" bucket within 96
blocks before falling through to findBlock — chunks with confirmed
trees are more likely to yield another.
stuck-incident body now includes journal byKind + last 12 scenario
entries so Pi can write a structural fix, not just a guard clause.
Architecturally: this is the foundation for "bot rewrites itself".
The proposals Pi now receives carry real signal about what was tried
and what's around, instead of a single snapshot in isolation.
10 new tests (world-journal × 5, scenario-memory × 5). npm test 134/134.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
@@ -0,0 +1,149 @@
|
||||
// Scenario memory: the bot remembers what skill it tried in what kind of
|
||||
// situation and how that worked out. Two consumers:
|
||||
//
|
||||
// 1. The scheduler skips a skill when (skillId, situationHash) showed
|
||||
// ≥N failures within MEMORY_WINDOW_MS — no point retrying the same
|
||||
// thing in the same context. The cooldown is shorter than the
|
||||
// per-skill backoff so the bot can come back later, but the hash
|
||||
// includes the current biome + time-of-day + nearby-block-types so
|
||||
// "I'm in a different place now" un-locks the skill automatically.
|
||||
//
|
||||
// 2. Pi-side proposals (stuck-incident) get a tail of recent entries
|
||||
// so the LLM can see what's been tried and reason about it
|
||||
// structurally instead of guessing.
|
||||
//
|
||||
// Persisted at state/<host>/scenarios.jsonl, append-only. Loaded into a
|
||||
// circular in-memory buffer for fast lookups. Old entries pruned by age.
|
||||
|
||||
import fs from "node:fs";
|
||||
import path from "node:path";
|
||||
import { stateDir } from "./config.js";
|
||||
|
||||
const SCENARIO_PATH = path.join(stateDir, "scenarios.jsonl");
|
||||
const MAX_LINES = 5_000;
|
||||
const TRIM_TARGET = 3_500;
|
||||
const MEMORY_WINDOW_MS = 30 * 60_000; // 30 min — recent enough to matter
|
||||
|
||||
function ensureDir() {
|
||||
try { fs.mkdirSync(stateDir, { recursive: true }); } catch {}
|
||||
}
|
||||
|
||||
// Cheap structural hash of the situation the bot is acting in: bucketed
|
||||
// position (16×16×8 cell), biome (if known), day/night, food bucket,
|
||||
// inventory keys (presence only, not counts), closest hostile type. The
|
||||
// hash is intentionally coarse — we want "same kind of place + same kind
|
||||
// of state" to match, not "exact (x,y,z) again".
|
||||
export function situationHash(snapshot) {
|
||||
if (!snapshot?.connected) return "disconnected";
|
||||
const p = snapshot.position ?? { x: 0, y: 0, z: 0 };
|
||||
const cx = Math.floor(p.x / 16);
|
||||
const cy = Math.floor(p.y / 8);
|
||||
const cz = Math.floor(p.z / 16);
|
||||
const day = snapshot.isDay ? "d" : "n";
|
||||
const food = snapshot.food === undefined ? "?" : snapshot.food >= 18 ? "F" : snapshot.food >= 12 ? "f" : "h";
|
||||
const hp = (snapshot.health ?? 20) >= 15 ? "H" : "L";
|
||||
const hostile = snapshot.closestHostile && snapshot.closestHostile.distance < 24
|
||||
? snapshot.closestHostile.name
|
||||
: "-";
|
||||
// Inventory keys, sorted — we lose counts but keep "what kind of stuff do
|
||||
// I have". Limited to first 10 names for hash stability.
|
||||
const invKeys = Object.keys(snapshot.inventory ?? {})
|
||||
.sort()
|
||||
.slice(0, 10)
|
||||
.join(",") || "-";
|
||||
return `${cx},${cy},${cz}|${day}|${food}|${hp}|host:${hostile}|inv:${invKeys}`;
|
||||
}
|
||||
|
||||
function loadScenarios() {
|
||||
const entries = [];
|
||||
let raw = "";
|
||||
try { raw = fs.readFileSync(SCENARIO_PATH, "utf8"); }
|
||||
catch (e) { if (e.code !== "ENOENT") return entries; return entries; }
|
||||
const now = Date.now();
|
||||
const lines = raw.split("\n").filter((l) => l.trim());
|
||||
for (const line of lines) {
|
||||
try {
|
||||
const entry = JSON.parse(line);
|
||||
const ts = Date.parse(entry.ts ?? "");
|
||||
if (!Number.isFinite(ts)) continue;
|
||||
if (now - ts > MEMORY_WINDOW_MS * 4) continue; // hard expire ×4 window
|
||||
entries.push({ ...entry, _ts: ts });
|
||||
} catch {}
|
||||
}
|
||||
return entries;
|
||||
}
|
||||
|
||||
function maybeTrim(linesSoFar) {
|
||||
if (linesSoFar <= MAX_LINES) return;
|
||||
try {
|
||||
const raw = fs.readFileSync(SCENARIO_PATH, "utf8");
|
||||
const lines = raw.split("\n").filter((l) => l.trim());
|
||||
const keep = lines.slice(-TRIM_TARGET).join("\n") + "\n";
|
||||
const tmp = `${SCENARIO_PATH}.tmp`;
|
||||
fs.writeFileSync(tmp, keep);
|
||||
fs.renameSync(tmp, SCENARIO_PATH);
|
||||
} catch {}
|
||||
}
|
||||
|
||||
export function createScenarioMemory({ failureThreshold = 3, windowMs = MEMORY_WINDOW_MS } = {}) {
|
||||
const entries = loadScenarios();
|
||||
let appended = entries.length;
|
||||
|
||||
function record({ skillId, situation, code, ok, detail }) {
|
||||
if (!skillId || !situation) return;
|
||||
const entry = {
|
||||
ts: new Date().toISOString(),
|
||||
skillId, situation, code, ok: !!ok,
|
||||
detail: detail ? String(detail).slice(0, 200) : null,
|
||||
_ts: Date.now(),
|
||||
};
|
||||
entries.push(entry);
|
||||
ensureDir();
|
||||
fs.appendFileSync(SCENARIO_PATH, JSON.stringify({ ...entry, _ts: undefined }) + "\n");
|
||||
appended++;
|
||||
if (appended % 200 === 0) maybeTrim(appended);
|
||||
}
|
||||
|
||||
function recentFailures({ skillId, situation, now = Date.now() }) {
|
||||
return entries.filter((e) =>
|
||||
e.skillId === skillId &&
|
||||
e.situation === situation &&
|
||||
!e.ok &&
|
||||
now - e._ts < windowMs
|
||||
);
|
||||
}
|
||||
|
||||
// Decision: "should I skip dispatching skill in this situation right now?"
|
||||
function shouldSkip({ skillId, situation, now = Date.now() }) {
|
||||
const fails = recentFailures({ skillId, situation, now });
|
||||
if (fails.length < failureThreshold) return false;
|
||||
// Also: have we succeeded with this same (skill, situation) lately? If
|
||||
// so, don't skip — the pattern may have changed.
|
||||
const recentOk = entries.find((e) =>
|
||||
e.skillId === skillId &&
|
||||
e.situation === situation &&
|
||||
e.ok &&
|
||||
now - e._ts < windowMs,
|
||||
);
|
||||
if (recentOk) return false;
|
||||
return true;
|
||||
}
|
||||
|
||||
// Pi-bound: tail of recent attempts as plain text (skill, situation, code, ok).
|
||||
function recentTailFor({ skillId, n = 10 } = {}) {
|
||||
const filtered = skillId ? entries.filter((e) => e.skillId === skillId) : entries;
|
||||
return filtered.slice(-n).map((e) => ({
|
||||
ts: e.ts,
|
||||
skillId: e.skillId,
|
||||
ok: e.ok,
|
||||
code: e.code,
|
||||
detail: e.detail,
|
||||
situationHashShort: (e.situation ?? "").slice(0, 60),
|
||||
}));
|
||||
}
|
||||
|
||||
function size() { return entries.length; }
|
||||
function clear() { entries.length = 0; try { fs.unlinkSync(SCENARIO_PATH); } catch {} }
|
||||
|
||||
return { record, recentFailures, shouldSkip, recentTailFor, size, clear };
|
||||
}
|
||||
Reference in New Issue
Block a user