First of three M3 cycles (Theme -> shell 2a -> Title). Palette.gd consts
as the single source of truth; game_theme.tres for flat-bordered UI;
shaders for the three signature surfaces. EB Garamond replaces
proprietary Georgia. Proof is a human-eyeballed showcase harness.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The aliveness POC (M0-M2) is proven, so redefine the project as the tactical
CRPG the /mockups describe. Charter: §1 reframed (Code of Conquest / The
Margreave), §10 combat = tactical + action-economy but GRIDLESS (tokens are
flavor, no positional rules), §15 = button-driven actions + free-text NPC
dialogue (Adjudicator reshaped to the optional free-text escape hatch), §16
adopts /mockups as the UI/systems bible, §17 rewritten to the game's scope.
Two conflicts flagged not silently decided (§18): damage determinism (§7 vs
mock ranges → M5) and class names (§8 vs mock callings → M4).
Roadmap: M0-M2 kept as the built engine; M3-M9 re-sequenced around the mock
screens (Theme+shell → creation → combat → dialogue → inventory/sheet → world
systems → framing/AI-flavor/persistence). Streaming stays shelved; v2 NPC
ideas preserved, re-homed to M6.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
High-fidelity HTML/CSS/JS design references for the tactical-CRPG direction —
title, character creation, character sheet, inventory, world map, quest log,
shop, pause/settings, dialogue, combat HUD, main exploration window. README is
the handoff spec (visual system, per-screen state models, Godot translation
notes). Not production code — the source of truth for the game's look + systems.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
M2 (prove aliveness) build-complete + live-proven; feel confirmed in playtest.
Active build focus is now M3 (make it a game), starting with the Adjudicator.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Flip considering-state to done+merged+live-confirmed. Strike the streaming
item — the considering-state reframes the wait and is tested/live; real
streaming's transport rework + mid-stream failure surface isn't worth it for
the POC. Streaming spec retained as a record (banner SHELVED). Includes the
human's folded-in further-down roadmap ideas.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Considering-state + transport timeout (M2 anti-freeze; streaming parked).
While a model call is in flight the client rotates authored in-character
'the Master considers…' lines (§13); HttpTransport now sets a request timeout
(35s, rejects <=0) so a hung call degrades to the fallback instead of spinning
forever. Human-confirmed working live, including the degrade path (API down →
fallback). 5 tasks, whole-branch review MERGE AS-IS. Rotation tuned to 3.0s.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
By-hand cold-start test showed 3 phrases before the response; 1.5x the
cadence so each line stays on screen a touch longer.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
5 TDD tasks: ConsideringPhrases (pure) + authored considering.json;
ProxyConfig.request_timeout_seconds (default 35s, reject <=0); HttpTransport
applies it (closes the infinite-hang gap); ConsideringIndicator node shim
(Timer rotates a Label); wire both harnesses around the await.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Pivot M2 anti-freeze from streaming to a lighter considering-state: a rotating
in-character 'the Master considers…' indicator (authored content, §13 spirit)
on the proven non-streaming loop, plus the HttpTransport request-timeout fix so
a hung call yields to the fallback instead of spinning forever. Meets §14's
anti-freeze goal by reframing the wait; near-zero new failure surface. Streaming
design PARKED (banner + roadmap ○), revisit if playtest shows the wait hurts.
Conscious §14 deviation recorded per §18.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Real token streaming to hide the ~2s call (§14). Proxy emits its own minimal
NDJSON (delta/done/error); /dm/narrate becomes a 200 stream (422 stays sync,
model errors become an in-stream frame). Client reworks to an HTTPClient
poll-loop StreamTransport behind a testable seam; hold-back tag filter;
keep-partial mid-stream degrade; folds in the HttpTransport timeout fix.
Narrator first; NPC fast-follow.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
NPC harness built a canon log with an empty player.luck_descriptor, which the
schema rejects (minLength 1) → /npc/speak 422. Build the player through
LogPlayer with a non-empty §7 fortune line. Human-confirmed working live.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The throwaway harness hand-built a minimal canon log and left luck_descriptor
as the default "", which fails the canon-log schema (minLength 1) → /npc/speak
422s and the harness silently degrades to fallback. Build the player through
LogPlayer with a non-empty §7 fortune line. Verified: request now 200s and Fenn
answers in-voice.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Bounded NPC conversation (/npc/speak), M2 core experiment. Server role is a
thin drop-in on the M1 pipeline (persona+knowledge server-side); client adds a
pure MoveValidator + MoveApplier over the proven DmService loop — validates
moves against live state, drops invalid, keeps prose (§6). Live-proven vs
qwen3.5 (Fenn in-voice, grounded in knowledge, no Luck leak). Live smoke caught
+ fixed two TagExtractor tag-format drifts. 18 tasks, all reviewed.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
M2 core experiment. Server owns persona+knowledge (spoilers/IP, no DB);
client owns live state and computes available_moves. Free text straight
to the NPC prompt (no Adjudicator). Sibling NpcService + pure MoveValidator;
moves validated against state, invalid dropped, prose always kept (§6).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
M2 client HTTP loop: the client posts its canon log to /dm/narrate, shows
the prose, harvests [FACT] via TagExtractor, and degrades to an authored
fallback (§13) on any failure. First end-to-end vertical slice, live-proven
against qwen3.5 through Docker. 67/67 client tests, whole-branch review
MERGE AS-IS.
Package api/prompts into the Docker image and give the Dockerized proxy a
route to a host-run Ollama. M1's narrator pipeline was proven via venv-on-
host; these were the two Docker-path gaps the client HTTP loop surfaced.
