Feature Cluster Dependency Matrix - M10-M11
See milestone-dependency-map.md for navigation.
| Cluster ID | Feature Cluster | Milestone | Depends On (Hard) | Depends On (Soft) | Canonical Docs | Decisions | Roadmap Phase | Gap Priority | Exit Gate | Parallelizable With | Risk Notes |
| M10.SDK.D038_CAMPAIGN_EDITOR | Campaign graph editor, intermissions, dialogue, named chars, testing tools | M10 | M9.SDK.D038_SCENARIO_EDITOR_CORE, M6.SP.FULL_RA_CAMPAIGNS | M9.SDK.GIT_VALIDATE_PROFILE_PLAYTEST | 09f-tools.md, modding/campaigns.md, 17-PLAYER-FLOW.md | D038, D065 | Phase 6b | — | Campaign authoring works for branching multi-mission campaigns | M10.GAME.D070_TEMPLATE_TOOLKIT | Scope explosion in intermission tooling |
| M10.SDK.D038_CHARACTER_PRESENTATION_OVERRIDES | Named-character presentation override convenience layer (voice/icon/portrait/sprite/palette/marker variants) with mission-scoped variant selection and preview | M10 | M10.SDK.D038_CAMPAIGN_EDITOR, M9.SDK.D040_ASSET_STUDIO | M9.SDK.EMBEDDED_AUTHORING_MANUAL, M10.GAME.D070_TEMPLATE_TOOLKIT | 09f-tools.md, modding/campaigns.md, 17-PLAYER-FLOW.md, 04-MODDING.md | D038, D021, D040, D068 | Phase 6b | — | Creators can define unique hero/operative readability (voice/skin/icon markers) without hiding gameplay changes in visual metadata; mission-level variant switching previews correctly | M10.GAME.MODE_TEMPLATES_MP_TOOLS, M10.MOD.D066_RA1_EXPORT_EXTENSIBILITY | P-Creator + P-Differentiator: keep gameplay-vs-presentation boundary explicit and avoid accidental compatibility/fingerprint confusion |
| M10.SDK.D038_CAMERA_TRIGGER_AUTHORING_ADVANCED | Advanced OFP-style camera-trigger authoring UI (shot graph, spline/anchor tools, trigger-context preview/simulate-fire, interrupt/fallback policy inspector) | M10 | M10.SDK.D038_CAMPAIGN_EDITOR, M6.UX.D038_TRIGGER_CAMERA_SCENES_BASELINE | M9.SDK.D040_ASSET_STUDIO, M9.SDK.GIT_VALIDATE_PROFILE_PLAYTEST | 09f-tools.md, 17-PLAYER-FLOW.md | D038, D065, D048 | Phase 6b | — | Designers can author and preview trigger-driven camera scenes with advanced shot tooling while still emitting normal trigger + Cinematic Sequence data | M10.UX.D038_RENDERED_CUTSCENE_DISPLAY_TARGETS, M11.VISUAL.D048_AND_RENDER_MOD_INFRA | Shot-graph/spline UX can sprawl; keep baseline trigger-camera property sheets sufficient for M6 content production |
| M10.UX.D038_RENDERED_CUTSCENE_DISPLAY_TARGETS | D038 rendered cutscene (Cinematic Sequence) advanced presentation targets: radar_comm / picture_in_picture capture surfaces, panel-safe framing preview, and fallback-aware validation | M10 | M10.SDK.D038_CAMPAIGN_EDITOR, M9.SDK.D040_ASSET_STUDIO, M6.SP.MEDIA_VARIANTS_AND_FALLBACKS | M9.SDK.GIT_VALIDATE_PROFILE_PLAYTEST | 09f-tools.md, 17-PLAYER-FLOW.md, 09d-gameplay.md | D038, D048, D065, D040 | Phase 6b | — | Rendered cutscenes can target fullscreen/world/radar/PiP with author-visible framing and validation, without breaking campaign fallback rules | M11.VISUAL.D048_AND_RENDER_MOD_INFRA | Capture-surface complexity and UI-framing drift can break readability; keep M6 baseline limited to world/fullscreen |
| M10.GAME.MODE_TEMPLATES_MP_TOOLS | Game mode templates + multiplayer scenario tooling + Game Master mode | M10 | M9.SDK.D038_SCENARIO_EDITOR_CORE, M7.NET.RANKED_MATCHMAKING (for MP semantics baseline) | M10.SDK.D038_CAMPAIGN_EDITOR | 09f-tools.md, 17-PLAYER-FLOW.md | D038 | Phase 6b | — | Multiple templates produce playable matches; MP scenario tooling works | M10.GAME.D070_TEMPLATE_TOOLKIT | Template sprawl before validation |
| M10.GAME.D070_TEMPLATE_TOOLKIT | Commander & SpecOps (D070) template toolkit and role-aware authoring/UX integration | M10 | M10.GAME.MODE_TEMPLATES_MP_TOOLS, M7.UX.REPORT_BLOCK_AVOID_REVIEW, M7.NET.CROSS_ENGINE_BRIDGE_AND_TRUST (policy awareness only) | M10.SDK.D038_CAMPAIGN_EDITOR | 09d-gameplay.md, 09f-tools.md, 09g-interaction.md, 17-PLAYER-FLOW.md | D070, D059, D065, D038, D066 | Phase 6b | — | D070 template validates, role HUDs and request lifecycle UX are wired | M10.GAME.EXPERIMENTAL_SPECOPS_SURVIVAL | Keep PvE-first and export limitations explicit |
| M10.GAME.D070_OPERATIONAL_MOMENTUM | Optional D070 pacing layer (Operational Momentum / “one more phase”) with agenda lanes, milestone rewards, and extraction-vs-stay prompts | M10 | M10.GAME.D070_TEMPLATE_TOOLKIT | M10.SDK.D038_CAMPAIGN_EDITOR, M10.GAME.EXPERIMENTAL_SPECOPS_SURVIVAL | 09d-gameplay.md, 09f-tools.md, modding/campaigns.md, 17-PLAYER-FLOW.md | D070, D038, D021, D065, D059 | Phase 6b/7 (prototype-first optional layer) | — | Optional pacing layer is validated without HUD overload; milestone rewards are explicit and can compose with Ops Prologue/Ops Campaign flags where authored | M10.GAME.EXPERIMENTAL_SPECOPS_SURVIVAL | P-Optional: timer walls, reward snowballing, or hidden mandatory chains can damage D070 readability if not tightly bounded |
| M10.GAME.EXPERIMENTAL_SPECOPS_SURVIVAL | Experimental Last Commando Standing / SpecOps Survival template | M10 | M10.GAME.D070_TEMPLATE_TOOLKIT | M10.GAME.MODE_TEMPLATES_MP_TOOLS | 09d-gameplay.md, 09f-tools.md, 17-PLAYER-FLOW.md | D070 | Phase 6b (experimental) | — | Experimental lobby/HUD/post-game surfaces exist and stay clearly labeled | — | Do not let this displace core template validation |
| M10.MOD.D066_RA1_EXPORT_EXTENSIBILITY | RA1 export + editor extensibility/plugin hardening + YAML/Lua extension tiers | M10 | M9.MOD.D066_OPENRA_EXPORT_CORE, M10.SDK.D038_CAMPAIGN_EDITOR, M9.COM.D049_FULL_WORKSHOP_CAS | — | 09c-modding.md, 09f-tools.md | D066, D038, D040, D049 | Phase 6b | — | RA1 export works for supported scenarios/campaign paths + editor extension system is safe | M10.SDK.LOCALIZATION_PLUGIN_HARDENING | API compatibility and capability manifests must be explicit |
| M10.SDK.RTL_BIDI_LOCALIZATION_WORKBENCH_PREVIEW | RTL/BiDi authoring-grade preview and validation in the Localization & Subtitle Workbench (mixed-script wrap/truncation, layout-direction preview, localized image/icon direction checks) | M10 | M10.SDK.D038_CAMPAIGN_EDITOR, M9.SDK.RTL_BASIC_EDITOR_UI_LAYOUT, M6.UX.RTL_BIDI_GAME_UI_BASELINE | M10.MOD.D066_RA1_EXPORT_EXTENSIBILITY | 09f-tools.md, 17-PLAYER-FLOW.md, 09c-modding.md | D038, D040, D065, D066 | Phase 6b | P3 localization tie-in | Creators can validate RTL/BiDi subtitles and UI strings (including directional assets/style variants) before publish/export | M10.SDK.LOCALIZATION_PLUGIN_HARDENING | If omitted, creators only discover RTL layout failures at runtime or after release |
| M10.SDK.LOCALIZATION_PLUGIN_HARDENING | Localization/subtitle workbench + editor plugin capability/version hardening + provenance release gating refinements | M10 | M9.SDK.D040_ASSET_STUDIO, M9.UX.RESOURCE_MANAGER_AND_PUBLISH_READINESS, M10.MOD.D066_RA1_EXPORT_EXTENSIBILITY, M10.SDK.RTL_BIDI_LOCALIZATION_WORKBENCH_PREVIEW | — | 09f-tools.md, 17-PLAYER-FLOW.md, 09c-modding.md | D040, D066, D068 | Phase 6b | P3 localization tie-in | Advanced authoring polish features (including RTL/BiDi preview/validation) land without cluttering simple mode | — | Keep simple/advanced separation intact |
| M10.COM.CREATOR_FEEDBACK_HELPFUL_RECOGNITION | Creator feedback inbox/review triage + helpful-mark workflow + profile-only reviewer recognition (badges/reputation/acknowledgements) | M10 | M9.COM.D049_FULL_WORKSHOP_CAS, M7.UX.POST_PLAY_FEEDBACK_PROMPTS, M7.NET.D052_SIGNED_CREDS_RESULTS | M11.COM.ECOSYSTEM_POLISH_GOVERNANCE, M11.COM.CONTRIBUTOR_POINTS_COSMETIC_REWARDS | 09e-community.md, 17-PLAYER-FLOW.md, 06-SECURITY.md | D049, D053, D031, D037, D052 | Phase 6b (with Phase 7 governance hardening) | — | Authors can triage feedback and mark reviews helpful; profile-only recognition is granted/revocable with clear trust labels and no gameplay effects | M11.COM.ECOSYSTEM_POLISH_GOVERNANCE, M11.COM.CONTRIBUTOR_POINTS_COSMETIC_REWARDS | P-Creator + P-Scale: collusion rings, alt-farming, and positivity-bias incentives require audit/revocation tooling |
| M11.COM.CONTRIBUTOR_POINTS_COSMETIC_REWARDS | Optional community-contribution points + cosmetic/profile reward catalog/redemption (non-tradable, non-gameplay) | M11 | M10.COM.CREATOR_FEEDBACK_HELPFUL_RECOGNITION, M11.COM.ECOSYSTEM_POLISH_GOVERNANCE | M11.PLAT.BROWSER_MOBILE_POLISH | 09e-community.md, 17-PLAYER-FLOW.md, 06-SECURITY.md | D049, D053, D037, D031, D052 | Phase 7 (optional ecosystem polish) | — | Points/redeemables are clearly profile-only, revocable, auditable, and cannot affect gameplay/ranked outcomes | — | P-Scale + P-Optional: reward-farming, inflation, and unclear wording can create abuse and player mistrust |
| M11.AI.D016_CONTENT_GENERATION | LLM mission/campaign/world-domination generation (built-in CPU floor + optional BYOLLM ceiling) | M11 | M10.SDK.D038_CAMPAIGN_EDITOR, M9.COM.D049_FULL_WORKSHOP_CAS | M6.UX.D065_ONBOARDING_COMMANDER_SCHOOL | 09f-tools.md, 04-MODDING.md, 17-PLAYER-FLOW.md | D016, D038 | Phase 7 | — | Optional generation works and outputs standard YAML/Lua | M11.AI.D047_PROMPT_STRATEGY | Must remain optional and fallback-safe |
| M11.AI.D047_PROMPT_STRATEGY | Four-tier provider management (built-in CPU / cloud OAuth / API key / local external), prompt strategy profiles, capability probing/evals | M11 | M11.AI.D016_CONTENT_GENERATION | M9.SDK.D040_ASSET_STUDIO, M2.SEC.CREDENTIAL_STORE_CORE | 09f-tools.md | D047, D016 | Phase 7 | — | Four-tier provider UX is reliable (Tier 1 built-in works offline after initial model-pack download; Tiers 2–4 BYOLLM connect/fail gracefully) with prompt strategy profiles | — | Local model template mismatch confusion |
| M11.AI.D057_SKILL_LIBRARY_EDITOR_ASSIST | LLM skill library + editor AI assistant tooling | M11 | M11.AI.D016_CONTENT_GENERATION, M11.AI.D047_PROMPT_STRATEGY, M9.SDK.D038_SCENARIO_EDITOR_CORE | M10.SDK.D038_CAMPAIGN_EDITOR | 09f-tools.md | D057, D016, D047, D038 | Phase 7 | — | AI assistance stays undoable, optional, and schema-grounded | — | Over-automation harming author control |
| M11.LLM.EXHIBITION_MODES | LLM exhibition matches (custom/local exhibition, prompt-coached, director-prompt showmatch, experimental BYO-LLM fight night), replay metadata/overlay, trust labels — never ranked (D055 hard exclusion) | M11 | M11.AI.D016_CONTENT_GENERATION, M11.AI.D047_PROMPT_STRATEGY, M7.NET.SPECTATOR_TOURNAMENT | M11.AI.D057_SKILL_LIBRARY_EDITOR_ASSIST | 09d-gameplay.md, 09f-tools.md, 09g-interaction.md | D073, D044, D047, D010, D055, D059, D071, D072 | Phase 7 | — | Three match surfaces operational (LLM Exhibition, Prompt-Coached, Director Prompt Showmatch); BYO-LLM fight night format functional; replay annotations with privacy controls; trust/mode badges visible | M11.COM.ECOSYSTEM_POLISH_GOVERNANCE | Ranked must never allow prompt-coached or director-prompt modes; prompts generate orders through normal pipeline only; vision scope (team/omniscient) must be disclosed; latency fairness is best-effort in BYO-LLM format |
| M11.AI.ML_TRAINING_PIPELINE | Replay-to-dataset conversion for custom AI model training — TrainingPair extraction (fog-filtered state + orders + outcome labels), Parquet export, SQLite training index, headless self-play generation, foreign replay corpus (D056), custom model integration paths (WASM AiStrategy, LlmProvider Tier 4, native Rust). Research-backed (spec in research/ml-training-pipeline-design.md) — not yet a settled decision; will be formalized as a D-number decision when implementation begins. | M11 | M2.CORE.SNAPSHOT_HASH_REPLAY_BASE, M2.COM.TELEMETRY_DB_FOUNDATION, M7.NET.CROSS_ENGINE_BRIDGE_AND_TRUST, M11.AI.D057_SKILL_LIBRARY_EDITOR_ASSIST | M11.LLM.EXHIBITION_MODES | research/ml-training-pipeline-design.md, 09d-gameplay.md | D044, D057, D031, D010, D056, D041 | Phase 7 | P-Optional | ic training generate/export/ingest CLI works; TrainingPair Parquet export validated; headless self-play produces usable datasets; at least one custom model integration path (WASM or Tier 4) demonstrated | — | Training data quality depends on replay quantity; custom models are user-provided (IC ships no trained weights); headless self-play requires stable sim before meaningful data generation; research spec — design details may evolve |
| M11.UX.D068_MACHINE_TRANSLATED_SUBTITLE_CC_FALLBACK | Optional D068 machine-translated subtitle/closed-caption fallback for missing media languages (clearly labeled, user opt-in, trust-tagged) | M11 | M6.SP.MEDIA_VARIANTS_AND_FALLBACKS, M9.UX.D049_MEDIA_LANGUAGE_CAPABILITY_METADATA_FILTERS, M10.SDK.LOCALIZATION_PLUGIN_HARDENING | M11.AI.D047_PROMPT_STRATEGY, M11.PLAT.BROWSER_MOBILE_POLISH | 09c-modding.md, 09e-community.md, 17-PLAYER-FLOW.md, 09f-tools.md | D068, D049, D038, D047, D037 | Phase 7 (optional) | — | Missing cutscene languages can fall back to machine-translated subtitles/CC when the player opts in, with explicit trust/source labels and no campaign progression breakage | — | P-Optional: mislabeled machine output, poor quality, or silent auto-enable can damage trust and localization expectations |
| M11.VISUAL.D048_AND_RENDER_MOD_INFRA | Switchable render modes + visual modding infrastructure (classic/HD/3D support, modder effects, and rendered-cutscene render-mode policy/fallback polish) | M11 | M1.CORE.RENDERER_SLICE, M9.SDK.D040_ASSET_STUDIO | M10.MOD.D066_RA1_EXPORT_EXTENSIBILITY, M10.UX.D038_RENDERED_CUTSCENE_DISPLAY_TARGETS | 09d-gameplay.md, 10-PERFORMANCE.md, 09a-foundation.md, 09f-tools.md | D048, D017, D015, D038 | Phase 7 | — | Optional render modes and visual infra exist without breaking low-end baseline, including author-declared rendered-cutscene render-mode preference/fallback behavior | — | Must preserve “no dedicated gaming GPU required” path |
| M11.PLAT.BROWSER_MOBILE_POLISH | Browser/mobile/Deck parity and platform-specific polish over existing abstractions (including RTL directionality polish across platform-specific surfaces) | M11 | M3.UX.FIRST_RUN_SETUP_AND_MAIN_MENU, M7.NET.TRACKING_BROWSER_DISCOVERY, M10.SDK.LOCALIZATION_PLUGIN_HARDENING | M11.VISUAL.D048_AND_RENDER_MOD_INFRA | 02-ARCHITECTURE.md, 09g-interaction.md, 17-PLAYER-FLOW.md | D069, D065, D059, D048 | Phase 7 | — | Platform variants remain obstacle-free and UX-consistent, including RTL directionality/mirroring behavior on browser/mobile/Deck UI surfaces | — | Platform-specific UX drift (including RTL layout divergence across platforms) |
| M11.COM.ECOSYSTEM_POLISH_GOVERNANCE | Governance tooling, community moderation polish, premium-content policy, creator ecosystem polish | M11 | M7.UX.REPORT_BLOCK_AVOID_REVIEW, M9.COM.D049_FULL_WORKSHOP_CAS | M10.GAME.MODE_TEMPLATES_MP_TOOLS | 09e-community.md, 17-PLAYER-FLOW.md | D037, D035, D046, D036 | Phase 7 | — | Governance/tooling/policy features mature after core platform trust exists | — | Avoid monetization/policy complexity before core community trust |