c910457817
* feat(v0.4.0): vNext — closed-loop world model + settlement contract
Implements the vNext architecture from the research doc: demote the noisy
multi-rail planner in favour of a closed loop (world truth → invariant check)
plus a single utility-driven goal authority.
L1 services (fix no_drop / silent pathfinder hang first):
- InventoryLedger: diff-based "did I actually get it" verifier; acquire-food
now confirms via ledger.gainedSince instead of the unreliable count/event.
- MotionService.gotoSafe: wall-clock timeout + progress watchdog +
path_update(noPath/timeout) → structured {reached|stuck|timeout|nopath}.
L3 plan — unify the three competing rails (curriculum/manifesto/storyline):
- Settlement Contract: ordered M0–M9 milestones, each invariant-checked
against an authoritative world view (early steps delegate to the proven
curriculum; late game adds farming).
- InvariantChecker + predicate library; GoalManager selects the lowest unmet
milestone via utility argmax (food-urgency preempts, DEPS-style).
- Wired into the scheduler: bot.js precomputes snapshot.contract; reflex.js
consumes it in place of the storyline rail. Manifesto L0 still preempts.
Eval + robustness:
- Village Score (single 0..1 metric) on the snapshot + TUI "build" line.
- survive.dig-in skill + dusk_dig_in mode (exposed at night, no bed → cover).
- approach_block helper (GoalNear + lookAt, avoids GoalLookAtBlock #341).
+28 new tests (450 total green). LLM remains entirely off the tick path.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
* feat(v0.4.0): finish vNext plan — anti-loop, skill-graph, worldDelta diff, flee→motion
Completes the remaining v0.4.0 plan items and one fix motivated by a live
in-game observation (flee hanging 30s against a persistent zombie).
- flee → MotionService.gotoSafe: structured {stuck|timeout|nopath} in ~4s with
a blind-retreat fallback, instead of the observed 30s pathfinder hang + 3
watchdog replans. Movements setup guarded so it is unit-testable.
- QW5 anti-loop (runtime/anti-loop.js): same skill failing >=3x in 5min →
30min blacklist (reflex shouldSkip) + one-shot improvement_request
(bot.js drainFired -> writeProposal).
- 4.1 closed-loop worldDelta: runSkill snapshots inventory before execute and
attaches the real delta (_invObserved) to every successful result; opt-in
skill.expectGain asserts the claimed gain or returns world_unchanged.
- 3.6 skill-graph (Plan4MC): declarative requires/produces for ~20 skills;
prerequisitesMet/canRun/runnableFrontier; GoalManager annotates suggestions
with blockedBy when prereqs are unmet.
+22 tests (472 total green). Live smoke confirmed dig-in works and no new
errors; flee loop is what this commit's flee migration addresses.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
---------
Co-authored-by: Yuriy Mayatnikov <mayatnikov@me.com>
Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
59 lines
2.6 KiB
JavaScript
59 lines
2.6 KiB
JavaScript
import { test } from "node:test";
|
|
import assert from "node:assert/strict";
|
|
|
|
import { createAntiLoop } from "./anti-loop.js";
|
|
|
|
test("fires after N failures in the window and blacklists the skill", () => {
|
|
const al = createAntiLoop({ windowMs: 60_000, threshold: 3, blacklistMs: 30_000 });
|
|
assert.equal(al.record({ skillId: "survive.flee", ok: false, now: 1000 }).fired, false);
|
|
assert.equal(al.record({ skillId: "survive.flee", ok: false, now: 2000 }).fired, false);
|
|
const third = al.record({ skillId: "survive.flee", ok: false, now: 3000 });
|
|
assert.equal(third.fired, true);
|
|
assert.equal(third.count, 3);
|
|
assert.equal(al.shouldSkip("survive.flee", null, 4000), true);
|
|
assert.equal(al.shouldSkip("survive.flee", null, 40_000), false); // blacklist expired
|
|
});
|
|
|
|
test("a success resets the fail streak", () => {
|
|
const al = createAntiLoop({ threshold: 3 });
|
|
al.record({ skillId: "gather.logs", ok: false, now: 1 });
|
|
al.record({ skillId: "gather.logs", ok: false, now: 2 });
|
|
al.record({ skillId: "gather.logs", ok: true, now: 3 });
|
|
const r = al.record({ skillId: "gather.logs", ok: false, now: 4 });
|
|
assert.equal(r.fired, false);
|
|
});
|
|
|
|
test("failures outside the window do not accumulate", () => {
|
|
const al = createAntiLoop({ windowMs: 1000, threshold: 3 });
|
|
al.record({ skillId: "s", ok: false, now: 0 });
|
|
al.record({ skillId: "s", ok: false, now: 500 });
|
|
const r = al.record({ skillId: "s", ok: false, now: 5000 }); // first two pruned
|
|
assert.equal(r.fired, false);
|
|
});
|
|
|
|
test("targetKey separates loops on different targets", () => {
|
|
const al = createAntiLoop({ threshold: 2 });
|
|
al.record({ skillId: "mine", ok: false, targetKey: "A", now: 1 });
|
|
const a2 = al.record({ skillId: "mine", ok: false, targetKey: "A", now: 2 });
|
|
assert.equal(a2.fired, true);
|
|
const b1 = al.record({ skillId: "mine", ok: false, targetKey: "B", now: 3 });
|
|
assert.equal(b1.fired, false); // different target, own streak
|
|
});
|
|
|
|
test("drainFired returns and clears the queue", () => {
|
|
const al = createAntiLoop({ threshold: 2 });
|
|
al.record({ skillId: "x", ok: false, now: 1 });
|
|
al.record({ skillId: "x", ok: false, now: 2 });
|
|
assert.equal(al.drainFired().length, 1);
|
|
assert.equal(al.drainFired().length, 0);
|
|
});
|
|
|
|
test("refire cooldown prevents immediate re-fire", () => {
|
|
const al = createAntiLoop({ threshold: 2, blacklistMs: 1000, refireCooldownMs: 100_000 });
|
|
al.record({ skillId: "x", ok: false, now: 1 });
|
|
assert.equal(al.record({ skillId: "x", ok: false, now: 2 }).fired, true);
|
|
// after blacklist expires, two more fails — within refire cooldown → no fire
|
|
al.record({ skillId: "x", ok: false, now: 2000 });
|
|
assert.equal(al.record({ skillId: "x", ok: false, now: 2100 }).fired, false);
|
|
});
|