Add silent original as townlet-noaudio.mp4

Replace demo with audio version; keep silent original as townlet-noaudio.mp4

Remove local absolute paths from CLAUDE.md notes

Update README.md

Update README.md

Tag Tiny Titan (<=4B) and Best Agent awards

Add demo video, blog, and social post links for submission

Disable CPU fallback: GPU-only on Spaces (CPU froze UI)

Non-blocking CPU fallback: mutex prevents UI freeze on lock contention

3 chars only; adaptive hog deadline (no more GPU task aborted)

README: model stack reflects single-model default + roster swap

All characters on nemotron-4b; one cold-load per fork

Hog GPU once acquired: duration=120, up to 12 sub-drains, 100s deadline

Dean-aligned Llama(): 5 kwargs only; hf_hub_download at module scope; poll 25s

Tier 1 safety: n_ctx 8192, TypeError guard on perf kwargs

Tier 1 perf: page-cache warm + n_ctx 6144 + q8_0 KV + mlock + poll 10s

@GPU (duration=120) + DECISIONS_PER_DRAIN=2 for tighter drains

Timestamped logs via game.log.tlog; explains visible drain pauses

Always probe GPU first; CPU is fallback per-tick, no lock

Add /tick_decisions_cpu fallthrough; broaden GPU error catch; JS GPU→CPU lock

Update README.md

CPU fallback when ZeroGPU quota exhausted; GPU/CPU chip in topbar

Move Info modal above script — fixes blank-UI from null DOM lookup

Add Cyra hero (saboteur); sharpen saboteur shell-as-weapon line

Fix Pause to halt drain; Spawn click feedback; Info modal with The Dot

Drop Flux 2; 4 starting chars; pin Ada=engineer Bram=diplomat

interpreter: per-fork interpreter so shell output reaches the UI (pid-mismatch detection)

docs: note @spaces.GPU fork semantics + the bugs they cause

interpreter: stop spurious 'mass death' flag (is_alive() unreliable across @spaces.GPU fork)

Named-place movement: atomic move_to, drop coord reasoning, 10/gather, per-archetype reward, default llama-3b, smart-mock updated

actions: strip surrounding quotes from move_to target (fixes "'cave'" loop)

actions: mine_coal/draw_water yield 10 units per action (was 1); tool descriptions updated so agent knows

actions: move_to short-circuits when already at destination so agent can mine/draw/post in the same decision

stream: strip <think> blocks + broaden meta-leak filter for Nemotron reasoning output

scheduler: idle drift skips characters at named places (fixes mine_coal not at cave) + strip <think> blocks from agent thought

stream: stop SoC meta-leak (few-shot exemplars + leak filter)

frontend: smooth char transition (0.8s linear) to match snapshot cadence

Recording prep: JS mutex, mixed models (Ada/Bram=Nemotron, rest=SmolLM2), drain=3, init=3, duration=90s, idle-drift, mock latency profile

frontend: slow tick_decisions poll 2s -> 20s (ZeroGPU quota)

Sync latest local snapshot

Default model: always Nemotron 3 Nano 4B, regardless of env

Mock: tighten SoC-detection so it doesn't swallow normal decisions

Trace: strip smolagents </code> marker from displayed thought

Frontend: revert panel order — SoC between Journal and Trace

CONTEXT: add Stream of consciousness glossary entry

Frontend: reorder panel — Goal/SoC/Trace promoted to top

Mock: templated SoC per archetype style for local testing

SoC: include in character's own next-decision perception prompt

SoC: hook into _decide + load/save with state.json

SoC: new game/stream.py — refresh every 10 decisions per char