Patches each link of the closed-loop chain (world-truth -> invariant ->
dispatch -> worldDelta -> milestone) so the contract can no longer advance
on a lie or stall on an unreachable milestone, and surfaces the result on
the monitor.
- Unblock M6_lighting: add gather.coal producer (mirrors gather.stone,
strict expectGain), register it, add the skill-graph node, and make
curriculum.shelter.torch two-step (gather coal -> craft torch). craft.torch
needed coal but no skill produced it, so the contract pinned on M6 forever
and starved M7/M8/M9. One curriculum edit fixes both rails.
- M8 no longer falsely satisfied: build-shelter records locations.shelter
only on real progress (setLocation moved below the no_progress guard).
- M2_bed made monotonic: sleepInBed writes locations.bed when the bot places
its own bed, so M2 stays met after the carried bed item is consumed.
- Observability: new ink-free tui/format.js helper renders per-tick
state / firing rail / active skill / blocked prerequisite / stall reason
in the monitor header (fields bot.js already computes but never showed).
- Harden the valid-skill-id guardrail: skillRegistryPrompt preserves the
"NEVER invent" footer when truncating (a growing registry was dropping it),
and fix the stale "tech-tree > autonomous" reflex chain in the escalation
prompt.
Tests: 494/494 passing; new build-shelter / invariants / format suites plus
skill-graph, curriculum, skill-registry cases. lint-patch clean. Deferred
audit findings recorded in plans/v0.4.0-vnext.md (local).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* 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>
mayatnikovandClaude Opus 4.8