diff --git a/docs/briefings/araminta.md b/docs/briefings/araminta.md
index 8f7201a63..a1acd77b0 100644
--- a/docs/briefings/araminta.md
+++ b/docs/briefings/araminta.md
@@ -29,7 +29,6 @@ Create Settled Reach visual style guide (color palette, UI patterns, visual gram
 Note: Image generation via /asset-gen costs money - always get Team Leader permission before generating. Needs to create a new style guide for this project (existing /asset-gen skill configured for a different project).
 
 ## Key Documents
-- DECISIONS.md - all confirmed decisions
-- DISCUSSION.md - active discussion round
+- decisions/ - domain-split decision files (see decisions/README.md for index)
 - docs/discussions/ - archived rounds
 - TEAM.md - team roster
diff --git a/docs/briefings/dudley.md b/docs/briefings/dudley.md
index ba77d1546..95790e907 100644
--- a/docs/briefings/dudley.md
+++ b/docs/briefings/dudley.md
@@ -8,19 +8,26 @@ The Settled Reach: top-down immersive sim, single-character perspective, asymmet
 STANDBY. This briefing will be populated when backend/engine implementation begins.
 
 ## Key Decisions for Your Domain
-- D-010: Four architectural principles - client-server, deterministic sim, information boundaries, input events
-- D-011: LOS shadowcasting implementation
-- D-012: Chunk-based map architecture
-- D-014: v0.1 map specification
-- D-020: Rust simulation server with bevy_ecs, MessagePack protocol to Godot client
-- D-021: Official title "The Settled Reach"
+Read `decisions/architecture.md` (primary) and `decisions/content.md` (NPC model). Key decisions:
+- **D-010:** Four architectural principles (client-server, info boundaries, deterministic sim, input events)
+- **D-020:** Rust simulation server with bevy_ecs, MessagePack protocol to Godot client
+- **D-024:** NPC ECS model (10 axes + CombatCapability component)
+- **D-026:** Simulation tier engine (Active/Background/State-saved/Ungenerated, timestamp LRU)
+- **D-030:** Testability architecture (hybrid Rust tests, deterministic replay critical)
+- **D-031:** Time system (10 ticks = 1 game-minute, 4 day phases)
 
-### Round 17 Decisions (Server-Side Core)
-- **D-024: NPC ECS model** - 10 axes as components (Want, Secret, Relationships 1-3, Tolerance, Routine, Information, Contentment, Personality, Tell, Skills). CombatCapability as optional component.
-- **D-025: Template instantiation system** - Single-ownership model with reference links carrying relationship metadata. NPCs owned by one template, referenced by others.
-- **D-026: Simulation tier engine** - Four tiers: Active (30-80 NPCs, full sim, 10-20 tps), Background (500-2K, state machines 1/game-min), State-saved (serialized structs), Ungenerated. Timestamp-based LRU eviction. bevy_ecs dynamic component add/remove for tier transitions.
-- **D-027: Access tier system + cargo/goods** - Content filtering by access tier (0-3). Cargo/goods system for smuggling mechanics.
+## Development Workflow
+See `docs/DEVOPS.md` for full procedures. Your key targets:
+- `make setup` — verify Rust toolchain (clippy + rustfmt)
+- `make build-server` — `cargo build` in `server/`
+- `make server` — `cargo run` in `server/`
+- `make test-server` — `cargo test` in `server/`
+- `make lint-server` — clippy (deny warnings) + rustfmt check
+- `make ci-server` — full server CI pipeline (lint → build → test)
+- Server code lives in `server/` — unit tests inline with `#[cfg(test)]`, integration tests in `server/tests/`
+- Cross-boundary IPC fixtures in `tests/`
 
 ## Key Documents
-- DECISIONS.md - all confirmed decisions
+- decisions/ - domain-split decision files (see decisions/README.md for index)
+- docs/DEVOPS.md - build, test, lint, CI procedures
 - docs/discussions/ - archived design rounds
diff --git a/docs/briefings/gestalt.md b/docs/briefings/gestalt.md
index e340d2e02..ac732b0b5 100644
--- a/docs/briefings/gestalt.md
+++ b/docs/briefings/gestalt.md
@@ -34,7 +34,6 @@ NPC ECS component architecture (10 axes + CombatCapability component). Social si
 **Iron rule for base builder (future):** CombatCapability component pattern must extend to other optional systems. No breaking the component model.
 
 ## Key Documents
-- DECISIONS.md - all confirmed decisions
-- DISCUSSION.md - active discussion round
+- decisions/ - domain-split decision files (see decisions/README.md for index)
 - docs/discussions/ - archived rounds
 - TEAM.md - team roster
diff --git a/docs/briefings/gore.md b/docs/briefings/gore.md
index da93cfc3f..87a20089d 100644
--- a/docs/briefings/gore.md
+++ b/docs/briefings/gore.md
@@ -27,7 +27,6 @@ Ensure vertical slice thematic spine is sound (both characters face "is this eno
 **Future flag:** Base builder DLC as thematic counter-offer to ascension (settling as active choice vs escape). Not now — but keep design space open.
 
 ## Key Documents
-- DECISIONS.md - all confirmed decisions
-- DISCUSSION.md - active discussion round
+- decisions/ - domain-split decision files (see decisions/README.md for index)
 - docs/discussions/ - archived rounds
 - TEAM.md - team roster
diff --git a/docs/briefings/hoshe.md b/docs/briefings/hoshe.md
index 9582e4bd6..88c444e0b 100644
--- a/docs/briefings/hoshe.md
+++ b/docs/briefings/hoshe.md
@@ -5,23 +5,11 @@ Last updated: 2026-02-10
 The Settled Reach: top-down immersive sim, single-character perspective, asymmetric information core mechanic, Rimworld-style storyteller. 29 confirmed decisions, 17 discussion rounds complete. Engine selected (D-020). Vertical slice defined (D-027).
 
 ## Decisions Relevant to Your Role
-- D-010: Information boundaries - testable? Can we verify information never leaks?
-- D-011: LOS shadowcasting - verify correctness of visibility calculations
-- D-012: Chunk loading - seamless? No visible pop-in or loading seams?
-- D-013: POI appear/disappear correctly? Diegetic insert triggers reliably?
-- D-015: Vision cone - correct blind spots? No information visible outside cone?
-- D-016: Monologue triggers - contextually appropriate? Right timing, right content?
-- D-017: Perception queries - each mode returns correct information for its type?
-- D-018: Three ranges working correctly? Sound quality degrades properly by distance?
-- D-020: MessagePack serialization - correct round-trip? No data loss at protocol boundary?
-- D-021: Official title "The Settled Reach"
-
-### Round 17 Decisions (Test Criteria)
-- **D-027: Vertical slice success criteria** - Concrete QA targets:
-  1. **30-minute breathing room** - Contamination activation must not trigger before player has 30 minutes of daily-life play (configurable for testing)
-  2. **Divergent playthroughs** - Same world seed, different character: verify relationships inverted, POIs differ, dialogue access varies
-  3. **Emotional attachment test** - Post-playthrough survey: can player name one NPC they felt conflicted about?
-  4. **Emergent discovery** - Observe→notice→follow→discover sequence must arise from systems (log-verifiable: no scripted reveals)
+Read `decisions/architecture.md` (D-010, D-020, D-030) and `decisions/perception.md` (D-011, D-017, D-018). Key decisions:
+- **D-010:** Information boundaries — testable? Can we verify information never leaks?
+- **D-020:** MessagePack serialization — correct round-trip? No data loss at protocol boundary?
+- **D-027:** Vertical slice success criteria (30-min runway, divergent playthroughs, emergent discovery)
+- **D-030:** Testability architecture (hybrid Rust tests, gdUnit4, three-layer IPC, CauseChain)
 
 ## Open Questions Assigned to You
 None assigned.
@@ -36,8 +24,20 @@ None assigned.
 
 All test plans due BEFORE implementation begins.
 
+## Development Workflow
+See `docs/DEVOPS.md` for full procedures. Your key targets:
+- `make test` — run all tests (server + client)
+- `make test-server` — Rust unit + integration tests
+- `make test-client` — gdUnit4 tests (headless runner pending)
+- `make lint` — run all linters
+- `make ci` — full CI pipeline locally (lint → build → test)
+- Server tests: `server/` (inline `#[cfg(test)]` + `tests/` directory per D-030)
+- Client tests: `client/` (gdUnit4 per D-030)
+- Cross-boundary tests: `tests/` (IPC fixtures, serialization round-trips)
+- Test caches: `.cache/` (gitignored, cleared by `make clean`)
+
 ## Key Documents
-- DECISIONS.md - all confirmed decisions
-- DISCUSSION.md - active discussion round
+- decisions/ - domain-split decision files (see decisions/README.md for index)
+- docs/DEVOPS.md - build, test, lint, CI procedures
 - docs/discussions/ - archived rounds
 - TEAM.md - team roster
diff --git a/docs/briefings/justine.md b/docs/briefings/justine.md
index 2b2a7a71f..fe4a7e133 100644
--- a/docs/briefings/justine.md
+++ b/docs/briefings/justine.md
@@ -8,14 +8,24 @@ The Settled Reach: top-down immersive sim, single-character perspective, asymmet
 STANDBY. This briefing will be populated when build/deployment pipeline work begins.
 
 ## Key Decisions for Your Domain
-- Q-007: Target platforms (open question - affects entire build pipeline)
-- D-019: Top-down primary view with future 3D cutscenes - build implications for asset pipelines and platform support
-- D-020: Godot 4 + Rust subprocess architecture
-- D-021: Official title "The Settled Reach"
+Read `decisions/architecture.md` (D-020, D-026, D-030) and `decisions/questions.md` (Q-007). Key decisions:
+- **Q-007:** Target platforms (open question — affects entire build pipeline)
+- **D-020:** Godot 4 + Rust subprocess architecture
+- **D-026:** State-saved NPC population — save game serializes 10K+ NPCs (~1-2KB each, ~10-20MB)
+- **D-030:** Testability architecture (CI pipeline structure, test runner tooling)
 
-### Round 17 Decisions (Save/Load Implications)
-- **D-026: State-saved NPC population** - Save game must serialize 10,000+ NPCs as frozen structs (~1-2KB each). Estimated save file size: ~10-20MB for NPC state alone. Compression strategy needed. State-saved NPCs reactivate on player return (~2-5ms per NPC).
+## Development Workflow
+See `docs/DEVOPS.md` for full procedures. You own the release pipeline.
+- `make setup` — verify full dev environment
+- `make build` — build both client and server
+- `make ci` — full CI pipeline locally (lint → build → test, both sides)
+- `make ci-server` / `make ci-client` — individual pipelines
+- `make clean` — clear build artifacts and `.cache/`
+- CI config: `.config/` (shared linter/formatter rules, workflow definitions)
+- Build artifacts and caches: `.cache/` (gitignored)
+- Tooling scripts: `tooling/` (asset pipelines, build helpers)
 
 ## Key Documents
-- DECISIONS.md - all confirmed decisions
+- decisions/ - domain-split decision files (see decisions/README.md for index)
+- docs/DEVOPS.md - build, test, lint, CI procedures
 - docs/discussions/ - archived design rounds
diff --git a/docs/briefings/mellanie.md b/docs/briefings/mellanie.md
index c736d4b22..022a3b705 100644
--- a/docs/briefings/mellanie.md
+++ b/docs/briefings/mellanie.md
@@ -43,6 +43,6 @@ The Settled Reach: top-down immersive sim, single-character perspective, asymmet
 **Critical tool:** Tyre's line previewer CLI (weekend build) — authoring tool and regression test harness.
 
 ## Key Documents
-- DECISIONS.md - all confirmed decisions
+- decisions/ - domain-split decision files (see decisions/README.md for index)
 - docs/discussions/ - archived design rounds
 - docs/briefings/miri.md - cultural generation guide (forthcoming)
diff --git a/docs/briefings/miri.md b/docs/briefings/miri.md
index 518d31891..35a178517 100644
--- a/docs/briefings/miri.md
+++ b/docs/briefings/miri.md
@@ -31,7 +31,6 @@ The Settled Reach: top-down immersive sim, single-character perspective, asymmet
 Also: ground both vertical slice characters (smuggler and detective) in setting-consistent cultural contexts.
 
 ## Key Documents
-- DECISIONS.md - all confirmed decisions
-- DISCUSSION.md - active discussion round
+- decisions/ - domain-split decision files (see decisions/README.md for index)
 - docs/discussions/ - archived rounds
 - TEAM.md - team roster
diff --git a/docs/briefings/nigel.md b/docs/briefings/nigel.md
index 0c398d242..db96c6819 100644
--- a/docs/briefings/nigel.md
+++ b/docs/briefings/nigel.md
@@ -26,7 +26,6 @@ Access tier system design (Tier 0-3 content pools). Seed-based population split
 **Key replayability multiplier:** Access tiers make Tier 1 conspiracy pool variable across playthroughs, not just within one game.
 
 ## Key Documents
-- DECISIONS.md - all confirmed decisions
-- DISCUSSION.md - active discussion round
+- decisions/ - domain-split decision files (see decisions/README.md for index)
 - docs/discussions/ - archived rounds
 - TEAM.md - team roster
diff --git a/docs/briefings/oscar.md b/docs/briefings/oscar.md
index c129ae74d..35dbef0c4 100644
--- a/docs/briefings/oscar.md
+++ b/docs/briefings/oscar.md
@@ -8,14 +8,23 @@ The Settled Reach: top-down immersive sim, single-character perspective, asymmet
 STANDBY. This briefing will be populated when networking/multiplayer implementation begins.
 
 ## Key Decisions for Your Domain
-- D-009: Multiplayer strategy (design for it, build single-player first)
-- D-010: Four architectural principles - especially client-server architecture and deterministic simulation
-- D-020: Subprocess/IPC architecture — local socket for single-player, TCP for multiplayer (same protocol)
-- D-021: Official title "The Settled Reach"
+Read `decisions/architecture.md` (primary). Key decisions:
+- **D-009:** Multiplayer strategy (design for it, build single-player first)
+- **D-010:** Four architectural principles (client-server, info boundaries, deterministic sim)
+- **D-020:** Subprocess/IPC architecture — local socket for single-player, TCP for multiplayer
+- **D-026:** Simulation tier sync — deterministic eviction across all observers for future multiplayer
 
-### Round 17 Decisions (Multiplayer Implications)
-- **D-026: Simulation tier sync** - For future multiplayer: how do we synchronize State-saved NPC reactivation across clients? Background tier state machines run server-side — clients receive snapshots. Timestamp-based eviction must be deterministic across all observers.
+## Development Workflow
+See `docs/DEVOPS.md` for full procedures. Your key targets:
+- `make setup` — verify Rust toolchain and Godot
+- `make build` — build both client and server
+- `make test` — run all tests (IPC round-trip tests critical for your domain)
+- `make ci` — full CI pipeline locally
+- Server code: `server/` (Rust, networking layer)
+- Client code: `client/` (Godot, network client)
+- Cross-boundary tests: `tests/` (IPC serialization fixtures, protocol regression)
 
 ## Key Documents
-- DECISIONS.md - all confirmed decisions
+- decisions/ - domain-split decision files (see decisions/README.md for index)
+- docs/DEVOPS.md - build, test, lint, CI procedures
 - docs/discussions/ - archived design rounds
diff --git a/docs/briefings/ozzie.md b/docs/briefings/ozzie.md
index 7b3e2c55b..6ae0398d1 100644
--- a/docs/briefings/ozzie.md
+++ b/docs/briefings/ozzie.md
@@ -23,7 +23,6 @@ None assigned.
 Vertical slice playtest evaluation. Champion the 30-minute breathing room success criterion. Verify both playthroughs (smuggler and detective) feel fundamentally different. Ensure social interaction moment-to-moment rewards match investigation moments. Guard the life-sim substrate from being sacrificed to conspiracy pacing.
 
 ## Key Documents
-- DECISIONS.md - all confirmed decisions
-- DISCUSSION.md - active discussion round
+- decisions/ - domain-split decision files (see decisions/README.md for index)
 - docs/discussions/ - archived rounds
 - TEAM.md - team roster
diff --git a/docs/briefings/paula.md b/docs/briefings/paula.md
index e7c2c8309..23374cfb2 100644
--- a/docs/briefings/paula.md
+++ b/docs/briefings/paula.md
@@ -24,7 +24,6 @@ The Settled Reach: top-down immersive sim, single-character perspective, asymmet
 NPC triangle authoring for vertical slice templates (workplace, bar, smuggling ring). Define cross-template relationship metadata schema for reference links. Design vertical slice character rosters (smuggler + detective perspectives on same NPCs). Map dialogue layers to NPC axes for systemic line selection. Social gameplay loop design — mundane triangles as standalone content.
 
 ## Key Documents
-- DECISIONS.md - all confirmed decisions
-- DISCUSSION.md - active discussion round
+- decisions/ - domain-split decision files (see decisions/README.md for index)
 - docs/discussions/ - archived rounds
 - TEAM.md - team roster
diff --git a/docs/briefings/qatux.md b/docs/briefings/qatux.md
index 160c5dab3..bbe3b5a55 100644
--- a/docs/briefings/qatux.md
+++ b/docs/briefings/qatux.md
@@ -34,7 +34,7 @@ None assigned directly.
 6. Verify briefing freshness against DECISIONS.md
 
 ## Key Documents
-- DECISIONS.md - all confirmed decisions
+- decisions/ - domain-split decision files (see decisions/README.md for index)
 - docs/discussions/ - archived rounds (including active rounds per D-022)
 - docs/briefings/ - agent briefing files (you maintain these)
 - TEAM.md - team roster
diff --git a/docs/briefings/si.md b/docs/briefings/si.md
index af56794ec..1e9e4a3e0 100644
--- a/docs/briefings/si.md
+++ b/docs/briefings/si.md
@@ -5,16 +5,12 @@ Last updated: 2026-02-10
 The Settled Reach: top-down immersive sim, single-character perspective, asymmetric information core mechanic, Rimworld-style storyteller. 29 confirmed decisions, 17 discussion rounds complete. Engine selected (D-020). Vertical slice defined (D-027). Next major workstream: vertical slice implementation.
 
 ## Decisions Relevant to Your Role
-All decisions (D-001 through D-029) - manages tickets derived from confirmed decisions.
-
-### Round 17 Summary (Sprint Planning Context)
-- **D-023:** Three-tier content architecture (Tier 1 authored modules, Tier 2 templated, Tier 3 procedural)
-- **D-024:** NPC generation model (10 axes + CombatCapability component)
-- **D-025:** Social site as atomic template unit (4-8 NPCs, 15-40 tiles, single-ownership + reference links)
-- **D-026:** Simulation tiers (Active/Background/State-saved/Ungenerated with timestamp eviction)
-- **D-027:** Vertical slice = smuggler + detective, station district, 3 templates, 30-min success criterion
-- **D-028:** Dialogue system (tagged pools, 4 relational layers, trait modifiers, template content packs)
-- **D-029:** Population entanglement ratio (30/50/20, variable per seed)
+All decisions across all domains. Read `decisions/README.md` for the index. Query decisions DB for sprint planning:
+```bash
+make decisions-coverage   # Decision-to-ticket coverage by domain
+make decisions-orphan     # Decisions without implementing tickets
+make decisions-active     # List all active decisions
+```
 
 ## Open Questions Assigned to You
 All open questions - tracks as tickets for assignment and progress monitoring.
@@ -29,8 +25,16 @@ Create sprint plan for vertical slice implementation (D-027). Break D-024 throug
 
 Vertical slice is the next major workstream — coordinate with Team Leader on sequencing and resourcing.
 
+## Development Workflow
+See `docs/DEVOPS.md` for full procedures. Key awareness for sprint planning:
+- `make ci` must pass before any PR merges — definition of done includes CI green
+- `make ci-server` = lint-server → build-server → test-server
+- `make ci-client` = lint-client → build-client → test-client
+- Server work lives in `server/`, client work in `client/`, cross-boundary tests in `tests/`
+- Hoshe's test plans target these directories per D-030 testing architecture
+
 ## Key Documents
-- DECISIONS.md - all confirmed decisions
-- DISCUSSION.md - active discussion round
+- decisions/ - domain-split decision files (see decisions/README.md for index)
+- docs/DEVOPS.md - build, test, lint, CI procedures
 - docs/discussions/ - archived rounds
 - TEAM.md - team roster
diff --git a/docs/briefings/stig.md b/docs/briefings/stig.md
index 1f0525732..70da8817f 100644
--- a/docs/briefings/stig.md
+++ b/docs/briefings/stig.md
@@ -8,23 +8,25 @@ The Settled Reach: top-down immersive sim, single-character perspective, asymmet
 STANDBY. This briefing will be populated when UI implementation begins.
 
 ## Key Decisions for Your Domain
-- D-013: Diegetic insert/POI navigation system
-- D-015: Camera locked to character, vision cone
-- D-016: Internal monologue text display
-- D-017: Perception mode overlays
-- D-018: Sound visual indicators (medium range)
-- D-019: Top-down primary view
-- D-020: Godot 4 client (GDScript)
-- D-021: Official title "The Settled Reach"
+Read `decisions/perception.md` (primary) and `decisions/scope.md` (D-013, D-014). Key decisions:
+- **D-013:** Diegetic insert/POI navigation system
+- **D-015:** Camera locked to character, vision cone
+- **D-016:** Internal monologue text display
+- **D-017:** Perception mode overlays
+- **D-019:** Top-down primary view, Godot 4 client (GDScript)
+- **D-028:** Dialogue UI — four relational layers (access tiers, history, trust-gated, unprompted)
 
-### Round 17 Decisions (UI Implications)
-- **D-028: Dialogue UI layers** - Four relational layers:
-  - Access tier indicators (insider/outsider/authority/peer)
-  - Relationship history feedback (trust/rapport display)
-  - Trust-gated dialogue options (locked topics until threshold met)
-  - Unprompted disclosure notifications (NPC volunteers information)
-- **D-027: Character-filtered insert UI** - Smuggler sees insider access, detective sees authority access. Same UI system, different filtered POI content.
+## Development Workflow
+See `docs/DEVOPS.md` for full procedures. Your key targets:
+- `make setup` — verify Godot is installed
+- `make client` — launch Godot with `client/` project
+- `make test-client` — gdUnit4 tests (headless runner pending)
+- `make lint-client` — gdlint/gdformat (pending setup)
+- `make ci-client` — full client CI pipeline (lint → build → test)
+- Client code lives in `client/` — scenes, GDScript, assets
+- UI tests use gdUnit4 (D-030)
 
 ## Key Documents
-- DECISIONS.md - all confirmed decisions
+- decisions/ - domain-split decision files (see decisions/README.md for index)
+- docs/DEVOPS.md - build, test, lint, CI procedures
 - docs/discussions/ - archived design rounds
diff --git a/docs/briefings/tiger.md b/docs/briefings/tiger.md
index e08b5a902..de3396d84 100644
--- a/docs/briefings/tiger.md
+++ b/docs/briefings/tiger.md
@@ -16,5 +16,5 @@ STANDBY. This briefing will be populated when localization work begins.
 - **D-028: Tagged line pool dialogue system** - Localization advantages: each line is tagged with context metadata (access tier, relationship level, trust threshold, mood). Translators receive rich context. Lines are pooled, not tree-branched — translation scope is predictable. Trait modifiers apply post-translation (base line + transformation guide = variants in target language).
 
 ## Key Documents
-- DECISIONS.md - all confirmed decisions
+- decisions/ - domain-split decision files (see decisions/README.md for index)
 - docs/discussions/ - archived design rounds
diff --git a/docs/briefings/troblum.md b/docs/briefings/troblum.md
index 4f495ecb4..e4a3ad929 100644
--- a/docs/briefings/troblum.md
+++ b/docs/briefings/troblum.md
@@ -24,7 +24,6 @@ The Settled Reach: top-down immersive sim, single-character perspective, asymmet
 Monitor D-026 simulation tier implementation for performance validation. Stress-test timestamp-based eviction against worst-case scenarios. Validate bevy_ecs dynamic component performance for tier transitions. Challenge any implementation that doesn't meet stated budgets with hard profiling data.
 
 ## Key Documents
-- DECISIONS.md - all confirmed decisions
-- DISCUSSION.md - active discussion round
+- decisions/ - domain-split decision files (see decisions/README.md for index)
 - docs/discussions/ - archived rounds
 - TEAM.md - team roster
diff --git a/docs/briefings/tyre.md b/docs/briefings/tyre.md
index ff9a3a5a3..b55ff9f0a 100644
--- a/docs/briefings/tyre.md
+++ b/docs/briefings/tyre.md
@@ -5,24 +5,12 @@ Last updated: 2026-02-10
 The Settled Reach: top-down immersive sim, single-character perspective, asymmetric information core mechanic, Rimworld-style storyteller. 29 confirmed decisions, 17 discussion rounds complete. Engine selected (D-020: Godot + Rust/bevy_ecs subprocess). Vertical slice defined (D-027).
 
 ## Decisions Relevant to Your Role
-- D-008: Z-levels, LOD, procedural generation - technical architecture implications
-- D-010: Four architectural principles - client-server, deterministic sim, information boundaries, input events
-- D-011: LOS shadowcasting implementation for fog of war
-- D-012: Chunk-based map architecture for streaming and generation
-- D-014: v0.1 map specification - full technical requirements
-- D-015: Camera lock to character - vision cone rendering and binaural audio hookup
-- D-017: Perception as observer queries - technical pattern for all perception modes
-- D-018: Sound propagation system - three-range model with wall occlusion
-- D-019: Top-down primary view with potential future 3D client
-- D-020: Godot 4 client + Rust/bevy_ecs server via subprocess/IPC, MessagePack serialization
-- D-021: Official title "The Settled Reach"
-
-### Round 17 Decisions (Technical Architecture)
-- **D-024: CombatCapability as ECS component** - Optional component triggered by combat-trained skill tag. Pattern applies to all optional systems. Follows D-017 perception mode model (capabilities as components, not monolithic state).
-- **D-025: Single-ownership template model with reference links** - NPCs owned by one template, referenced by others. Reference links carry relationship metadata. Prevents lifecycle conflicts.
-- **D-026: Simulation tier budgets confirmed** - Your numbers: Active (30-80 NPCs, 10-20 tps), Background (500-2K NPCs, state machines 1/game-min, 4 machines each), State-saved (~1-2KB serialized structs), Ungenerated (doesn't exist). Timestamp-based LRU eviction. bevy_ecs dynamic components for tier transitions.
-- **D-027: Vertical slice systems** - Access tier system + cargo/goods system needed for smuggling ring mechanics. Two characters = ~20% more effort, no new architecture (client-server already supports multi-observer).
-- **D-028: Line previewer CLI** - Weekend build. Authoring tool and regression test harness. Takes template YAML → NPC instances → trait modifiers → outputs generated line pools with metadata.
+Read `decisions/architecture.md` (primary) and `decisions/perception.md` (secondary). Key decisions:
+- **D-010:** Four architectural principles (client-server, info boundaries, deterministic sim, input events)
+- **D-020:** Godot client + Rust/bevy_ecs server via subprocess/IPC, MessagePack serialization
+- **D-026:** Simulation tier budgets (Active 30-80 NPCs, Background 500-2K, State-saved 10K+)
+- **D-030:** Testability architecture (hybrid test org, gdUnit4, CauseChain, three-layer IPC testing)
+- **D-031:** Time system (10 ticks = 1 game-minute, 4 day phases)
 
 ## Open Questions Assigned to You
 - Q-001: Engine selection (LEAD, CRITICAL - next priority)
@@ -41,8 +29,19 @@ Vertical slice technical implementation:
 
 **Future flag:** Base builder DLC requires data model stub now: `modifications: Vec<Modification>` on relevant entities. Add to schema early.
 
+## Development Workflow
+See `docs/DEVOPS.md` for full procedures. You own the technical build pipeline.
+- `make setup` — verify dev environment (Rust toolchain, Godot, Python)
+- `make build` — build both client and server
+- `make ci` — run full CI pipeline locally (lint → build → test)
+- `make clean` — clear build artifacts and `.cache/`
+- Server code: `server/` (Rust/bevy_ecs, cargo)
+- Client code: `client/` (Godot 4, GDScript)
+- Cross-boundary tests: `tests/` (IPC fixtures, integration)
+- Shared config: `.config/` (linters, formatters)
+
 ## Key Documents
-- DECISIONS.md - all confirmed decisions
-- DISCUSSION.md - active discussion round
+- decisions/ - domain-split decision files (see decisions/README.md for index)
+- docs/DEVOPS.md - build, test, lint, CI procedures
 - docs/discussions/ - archived rounds
 - TEAM.md - team roster