10 tickets across server (6), client (3), copy (1).
Capstone sprint for v0.1 — everything converges on a full
playthrough from main menu through storyteller activation.
Closed stale epics: #38, #369, #455, #575, #596.
Sprint goal: wire TriangleActivated into player-visible signal,
thread character archetype through session lifecycle.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Review fixes:
- Terminal: fix sightlines comment col 22 → col 28 (actual window position)
- Last Shift: fix door comment col 21 → col 22, add spatial features
(bar counter W-stubs cols 1-2, table clusters, card table, ticker mount)
- Maintenance corridors: extend transition corridor from 26 to 40 tiles
per D-093 spec (grid now 58x6)
- Gate ground: expand customs zone from 8 to 10 rows per D-093 spec,
correct freight (20 tiles west) / ped (10 tiles east) layout with
6-tile corridor between (grid now 40x34)
- Gate gallery: add staircase entrance tile (col 1 row 0 = R)
- District description: update from "Three social sites" to five locations
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replace worktree-centric language with team-centric framing across
CLAUDE.md and skills. Agents now identify by $WORKTREE_TEAM env var
instead of resolving git internals. This prevents agents from
following .git pointers back to the main repo and crossing boundaries.
- CLAUDE.md: rename section to "Team boundaries", reference $WORKTREE_TEAM
- sprint-start: add TEAM BOUNDARY rule to agent spawn prompt
- sprint-plan: replace "worktree-relative paths" with "relative paths only"
- pr-review: replace "worktree" with "team directory", note cross-dir
reading is a main-team privilege only
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Complete tile data for all five locations in the transit district:
- The Terminal (44x28, z=1): logistics hub with supervisor office LOS window
- The Last Shift (34x22, z=1): bar with corner booth and back room alley exit
- Maintenance Corridors (44x6, z=0): restricted storage, hatch room, transition corridor
- Gate Ground (40x32, z=1): aperture chamber, staging, customs lanes, concourse
- Gate Gallery (32x10, z=2): Commission-only observation gallery
Gate corridor split into two files (gate-ground, gate-gallery) because
the Location struct supports only one tile_bounds per file.
Updated district.yaml locations list with gate-ground and gate-gallery.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add D-088 Overlay pause/unpause signals to DebugConsole, wire in main.gd
so sim does not advance while typing debug commands
- Settings dialog reads live DebugConsole.is_enabled() instead of ConfigFile
directly, preventing checkbox/state divergence
- append_response respects disabled state — no auto-open when user disabled
console via settings
- tp command warns on invalid z value instead of silently defaulting to 0
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Server #580 bumped to 18 for debug_response field. Client was still
at 17, causing every snapshot to be rejected — game unplayable.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
4 new tests in test_fog_sprint22.gd verify BoundaryWall tiles populate
boundary_positions (not visible_positions), get VIS_FORWARD without
EXP_VISIBLE, stay EXP_UNEXPLORED after leaving LOS, and clear on new
snapshot. Comment in tile_renderer.gd documents implicit rendering path.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- game_state.gd: add boundary_positions Dictionary field; BoundaryWall tiles from
visible_tiles go to boundary_positions instead of visible_positions — rendered by
tile_renderer but not tracked as explored fog memory
- fog_state.gd: update_from_state() writes VIS_FORWARD for boundary_positions so fog
lifts over margin wall content; boundary tiles excluded from exploration step so they
don't persist as EXP_EXPLORED when player turns away
- tile_renderer.gd: no changes needed — renders all visible_tiles by type, sector-agnostic
- test_fog_shader.gd: 4 new tests — boundary excluded from visible_positions, tracked in
boundary_positions, cleared each snapshot, fog lifts to VIS_FORWARD
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Same Sprite2D correction applied to test_client_p2.gd entity color tests
(Terrain, Player) — ColorRect was replaced with Sprite2D in entity_renderer.gd.
Minor comment clarification in test_rendering.gd rotation test.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Reset _npc_entity_colors/_npc_entity_id/_next_npc_color in _end_player_conversation()
to prevent palette exhaustion across long sessions with many unique NPCs
- Re-enforce contrast floor after passive desaturation (_enforce_contrast after _desaturate)
- Add TestDialogueSpeakerColors suite: palette allocation, entity reuse, reset, fallback
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Maintains Dict[entity_id → Color] in dialogue_box for player conversations.
On first encounter, assigns a round-robin palette color; reuses on subsequent lines.
Eliminates position-based name-hash coloring for player dialogue.
Changes:
- Add _npc_entity_colors dict, _npc_entity_id, _next_npc_color fields
- Add _assign_npc_color(entity_id) — registers palette color on first encounter
- show_dialogue: accept npc_entity_id param, register entity color
- append_line: optional speaker_entity_id/target_entity_id stored in log entries
- append_player_line: pass _npc_entity_id as target_entity_id
- append_dialogue_response: accept entity_id, register, pass to append_line
- _format_entry else branch: look up _npc_entity_colors before name-hash fallback
- main.gd: pass _last_dialogue_npc_id to show_dialogue, speaker_entity_id to append_dialogue_response
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Fixes 7 pre-existing failures in test_entity_renderer tests:
- ColorRect → Sprite2D cast; .color → .self_modulate for D-033 color checks
- Position offset: (TILE_SIZE-24)/2 → EntityRenderer.ENTITY_OFFSET_{X,Y} (0.0)
- Rotation accuracy: expected values updated for raw un-normalised Godot rotation
Also adds SoundIndicatorRenderer to global_script_class_cache.cfg so test_rendering.gd parses.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Add debug_response: None to all ObserverSnapshot constructors in
integration tests
- Bump PROTOCOL_VERSION assertion 17 → 18 in serialization tests
- Regenerate golden proof_room_tick_10.json (BoundaryWall tiles)
- Regenerate client msgpack fixtures for new snapshot fields
- Fix debug.rs resource optionality (Option<ResMut> for
ContaminationActive/EventQueue)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Debug and LOS tracks for Sprint 23 (#580, #584):
- DebugCommandKind enum with 10 variants (AdvanceTicks,
SkipToContamination, TeleportToPosition, ForceContaminationActivate,
InspectNpc, ListTriangles, ListPopulation, GetContaminationStatus,
TeleportToLocation, ForceTriangleActivation)
- DebugResponsePayload on ObserverSnapshot, handle_debug_commands
system gated by DebugEnabled resource
- PROTOCOL_VERSION bumped 17 → 18
- VisibilitySector::BoundaryWall variant — 1-tile wall margin beyond
LOS boundary included in visible_tiles (not exploration/memory)
- compute_boundary_walls() pass in NaturalVision after FOV+cone
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Storyteller chain for Sprint 23 (#570, #571, #572, #579):
- EngagementRecord component: per-NPC observation_time_ticks,
conversation_count, monologue_trigger_count — incremented by
perception, dialogue, and monologue systems
- MovementHistoryBuffer resource: ring buffer of player positions
over last 3000 ticks with npcs_copresent_in_window() query
- Lifecycle rules: single activation per session, no concurrency,
no cooldown, terminal resolution constants
- activation_pass() system: gate check, proximity query, engagement
scoring, unentangled-NPC routing, module selection, emits
TriangleActivatedEvent on 10-tick cadence after contamination
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Spatial chain for Sprint 23 (#576, #577, #578):
- TileKind enum (Floor/Wall/Void/Restricted) on WalkabilityMap with
set_tile_kind/tile_kind API, backward-compatible with existing
is_walkable/set_walkable
- Location YAML tile format: tiles as string arrays (F/W/V/R chars),
load_location_tiles() stamps tile data onto WalkabilityMap from
ContentStore on production startup
- Chunk streaming system: ChunkLoadRadius + ChunkStreamingCadence
resources, loads/unloads chunks around player position on cadence.
v0.1 radius covers full district (no streaming stutter)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Godot 4.6 OpenGL3 compatibility mode does not support 'return' in
fragment(). The early returns on lines 68 and 82 caused silent shader
compilation failure, making the fog overlay render as a no-op — the
root cause of the Sprint 22 fog regression.
Restructured to if/else-if/else chain preserving identical logic.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The 5-layer fog model's deep fog sub-zone (alpha 0.55-0.70) was designed
for the old peripheral sector. After #569 removed peripheral, tiles behind
the player dropped straight to deep fog — indistinguishable from unexplored
black over dark scene tiles.
D-015 is explicit: behind = light fog overlay, art preserved, just "not
fresh." There is no deep fog sub-zone.
Changes:
- Remove dual light/deep fog alpha — all explored tiles use alpha 0.25-0.35
- Add 5x5 Gaussian blur on exploration texture to soften staircase boundary
- Use blurred exploration value for the visual transition (raw for anti-bleed)
- Remove DARK_OVERLAY constant (zone_tint handles all explored color)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The merge of visual and client branches dropped the `tiles` variable
declaration. The client branch refactored bounds calculation to use
`visible_positions`, but the visual branch's zone tint loop still
iterates over `visible_tiles` for zone_id data.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Register StorytellerPlugin in main() and dump_schedule_graph() so
contamination system runs in production (critical, rounds 2+3)
- Add doc comment to z_bands_connected clarifying band indices vs
absolute z-levels (D-110)
- Add TODO on hardcoded modifications: vec![] in save_to_file
- Init ContaminationActive in minimal_world() test helper
- Replace ChaCha20Rng with SimRng in fuzzy_map tests (D-010)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Round 4 review: mark VIS_PERIPHERAL as deprecated (peripheral sector
removed in #569, constant retained for test compatibility). Fix spec
status table row from "peripheral sector" to "cone gradient".
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Round 3 review fixes — thorough spec cleanup:
- Remove visibility_sectors from data flow (peripheral removed in #569)
- Remove player_pos uniform (cone center implicit in visibility_tex)
- Update FogState pseudocode: remove sector step, add zone tint step
- Update lifecycle diagram to match single update_from_state() call
- Fix "5-layer fog" → "3-state fog" in Files to Create and impl notes
- Mark zone tint open question as resolved (Sprint 22, D-077)
- Document filter_nearest rationale on zone_tint_tex (D-073 hard zones)
- Note low-saturation tint is intentional per D-046 Hopper test
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Three save/load bugs fixed:
- ContaminationActive not persisted in SaveStateV1 — caused double-fire
of contamination pressure on reload after tick 300.
- Loaded triangle entities missing ActiveSim marker — made them
invisible to escalation and contamination systems after any load.
- Existing triangle entities not despawned before load — created
duplicates, doubling tension escalation per tick.
Also: HashSet → BTreeSet for D-010 compliance, defensive event queue
reset on load, and three regression tests for triangle roundtrip.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The constant was supposed to represent 30 game-minutes but the formula
was wrong (30 × 10 tps × 60s = 1800). Correct derivation: 30 minutes ×
TICKS_PER_GAME_MINUTE (10) = 300. Now uses the canonical constant
directly. Also fixes stale assertion message in integration test.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
GDScript int is i64 — when randi() returns a value with bit 31 set,
left-shifting by 32 sets bit 63, producing a negative i64. MessagePack
encodes this as a negative integer, which Rust rmp_serde rejects when
deserializing as u64, causing ~50% startup failure rate.
Fix: mask bit 31 before shifting in new_game() to cap entropy at 63
bits. Also mask the sign bit in _read_seed_file() to handle save files
written before this fix.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
GLSL sample comments now say ±0.05 / ±0.075 matching the code and
spec table. Implementation notes gradient radius updated from 6-8
to 3-4 tiles.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Remove the db/connectors → tooling/db/ symlink added in Sprint 21
(#274) and migrate all references to use tooling/db/ directly.
- Delete tracked symlink from db/connectors
- Remove duplicate db/connectors/* permission patterns from settings
- Update project-structure.md to reflect removal
- Move whatsinagame/static/db/connectors/ to whatsinagame/static/tooling/db/
- Update 20 whatsinagame template, skill, and test files
- Update comment references in client/tests/test_anti_tedium.gd
- Historical docs (old sprint briefings, changelog, discussions) left as-is
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The 7x7 Gaussian kernel (sigma 2.0) produces a 3-4 tile radius
gradient, not 6-8 tiles. Header comment now matches the function
comment and actual math.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Widen world_seed entropy from u32 to full u64 by combining two randi()
calls (Hoshe warning #1)
- Persist world_seed to save directory and restore on resume_game() so
loaded sessions maintain D-010 deterministic replay (Tyre warning #2)
- Constrain EntanglementConfig intrigue range based on flat value so
mundane_ratio stays within D-029 spec [45,55]% (both reviewers)
- Remove dead VIS_PERIPHERAL constant and _grow_bounds() method
- Update test_client_p1 peripheral test for forward-only simplification
- Fix misleading exp_fade shader comment (filter_nearest = hard step)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Sync fog-shader-spec.md with actual shader values after #563 tuning:
light fog 0.25-0.35 (was 0.26-0.34), deep fog 0.55-0.70 (was 0.54-0.70).
Pseudocode now uses symmetric noise remapping (noise*2-1)*amp to match
the shader. Added first-call guard comment on _tint_bytes in fog_state.gd.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
