Firestaff

Firestaff Completion Matrix

Last updated: 2026-10-24

This is a conservative 100-point parity-completion model. A point is counted only when the repo has source, runtime, or CI evidence for that criterion. completionPercent is therefore a verified evidence score, not optimism, effort spent, or pass count.

Public wording should stay honest: these numbers describe verified Firestaff evidence coverage, not a claim that every listed game is fully playable.

Current roll-up

The matrix currently contains 9 runtime targets and 41 recognised version profiles. A version profile is an asset-admission identity; it is not an independent parity implementation. The percentages below therefore remain target-level evidence scores. They must not be multiplied by the number of hashes or presented as 41 separately verified ports.

Game family Runtime targets Version profiles Mean runtime evidence Version-profile scope
DM1 2 16 64.5% PC 3.4 is the source-locked playable route; other profiles are catalogued/admitted where their data gates pass.
CSB 2 9 37.0% Shared CSB runtime slices; edition-specific launch/data boundaries are not full parity proof.
DM2 2 7 26.0% PC/FM Towns/Amiga/PC-98 profiles are recognised; authentic DOS, Amiga, FM Towns and Mac V1 start paths are verified, but live parity remains open.
DM Nexus 2 5 28.5% Saturn JP/extracted, V2, English fan and French fan profiles share the capture-gated Nexus route.
Theron’s Quest 1 4 72.0% JP Rev 1 CUE/BIN reaches the Track 02 runtime route, and the US CloneCD ZIP reaches direct and start-menu launch through its verified Track 02 slice. Level-transition consumer captures remain open.
Total 9 41 42.7% Unweighted mean of the 9 runtime targets: 384/900.

The family and total means are derived from the nine target scores in the machine-readable matrix. Version profiles are not independent parity targets, so their count is not used to weight the score.

The latest authentic Nexus Saturn captures remain evidence additions, not a score promotion: VDP1 command-to-DGN pixel/CLUT joins are verified for selected frames, while menu/FONT256 ownership, complete VDP1 scene ordering, HUD/VDP2 composition and SLEV/SAL/SDDRVS playback remain blocked.

Scoring criteria

Criterion Weight Meaning
reference_inventory 10 Original/source reference inventory exists and is source-locked for this target.
definition_matrix 10 Target-specific parity/completion matrix exists with clear acceptance labels.
launch_smoke 10 Target launches, blocks correctly, or renders through a repeatable repo gate/smoke path.
core_input_movement 15 Core input, movement/camera, command routing, and first viewport redraw path are source/runtime-gated.
viewport_ui_render 20 Viewport, HUD, panels, title/menu, palette, graphics decode, and UI render surfaces are source/runtime-gated.
gameplay_systems 15 Combat, creatures, items, spells, sensors, projectiles, save/load, lifecycle, and dungeon systems are original/source-gated.
audio_timing 10 Audio content, trigger cadence, animation/timing, input responsiveness, and music/SFX overlap are original/source-gated.
original_overlay_regression 10 Original runtime captures/overlays/regression gates prove pixel/behavior parity for representative states.

Current matrix

Target completionPercent Points Status Primary blockers
DM1 V1 78% 78/100 playable_verified_partial Representative original-vs-Firestaff overlay parity remains incomplete. Native archive startup is now separately proven for both the canonical PC 3.4 ZIP and the supplied nested DOS-EN layout; the missing scope is original runtime overlay/capture evidence, not archive ownership.
DM1 V2 51% 51/100 runtime_presentation_verified_partial V2.0/V2.1 runtime presentation smoke exists, but finished V2.2 real-art material/pixel gates and original-pairing evidence remain open.
CSB V1 43% 43/100 runtime_boundary_verified_partial CSB has authentic native startup/runtime and start-menu evidence on Atari ST, Amiga and FM Towns. Atari, Amiga and both FM Towns editions verify their complete initial input matrices from independent source sessions. The French preservation ZIP verifies its original protected STX sector ordering through the same route. End-to-end gameplay/render capture proof is still being hardened.
CSB V2 31% 31/100 presentation_verified_partial CSB V2 HUD/filter/smooth-movement probes are wired, but real CSB V1 gameplay parity and modern-art per-cell material proof remain open.
DM2 V1 36% 36/100 runtime_slices_verified_partial Authentic DOS, Amiga, FM Towns and Mac startup/runtime evidence joins door/button, projectile, creature, minimap, weather/timer, save-adjacent slices, combat damage pipeline, and movement execution chain. V1 dungeon/render/mechanics parity remains active work.
DM2 V2 16% 16/100 presentation_scaffold_partial V2 presentation scaffolds exist, but V1 parity base and finished V2 proof are not complete.
DM Nexus V1 37% 37/100 runtime_slices_verified_partial DMDF/DGN/data, BPK/BPX, font, save, and mechanics slices exist; real-asset end-to-end handoff and screen capture proof remain active work.
DM Nexus V2 20% 20/100 presentation_scaffold_partial V2 lighting/touch/upscale/presentation probes exist, but real V1 runtime coverage and V2 material proof remain incomplete.
Theron V1 72% 72/100 hw_config_cd_play_vdc_joypad_proven Original overlay regression (0/10) is the largest remaining gap. Real menu implementation needed.

Recent evidence incorporated

This refresh incorporates the post-2026-05-09 work now present on origin/main, including:

Area Evidence now counted
DM1 V1 Hall of Champions mirror/portrait runtime probes across many ordinals and routes; resurrect survival/load regression; chest runtime matrix; food/water warning tick; door-bash sound/no-open; projectile grate/pass-through; spell-rune repeat input; champion-panel disabled icon and mana repaint gates; movement/data-root CTest hardening.
DM1 V2 Runtime presentation smoke for V2.0/V2.1; source-route and side-by-side seed gates; HUD overlay/interaction gates; lighting/effects/field projectile VFX gates; selected-resolution and 4K input mapping; per-mode material-signature and renderer readback probes.
CSB V1 Authentic native Atari STX and nested-ZIP, including the supplied French preservation ZIP → protected STX route, Amiga ADF-archive, and English/Japanese FM Towns-CD startup paths now prove title → runtime and start-menu handoff. Atari, Amiga and FM Towns independently cover their complete initial input matrices. Additional evidence includes CSB save runtime boundary; command-chain/input/runtime tick slices; dungeon header/loader/world gates; utility/CMP import; DSA trigger; save import path; viewport source-lock slices and CustomBackgrounds coverage; hidden-code skip tables.
CSB V2 HUD overlay probe, smooth movement/runtime binding, phase-gate verification, filter config/chain/dispatch, texture upscale, V2.2 shape probes, and CTest registration of previously build-only probes.
DM2 V1 Authentic DOS MVE → SKULL → New Game CLI, Amiga title/runtime, FM Towns title/New Game/champion/runtime, and Mac retail boot; the separately supplied Mac First Chapter demo is fail-closed and cannot select a retail sibling; door/button toggle boundary; minimap level-transition/save-load gate; weather/timer save round-trip; projectile-vs-creature collision; creature death/drop observer; pressure plate and trigger gates; focused combat/shop probes.
Nexus V1 DMDF palette/string bounds, DGN malformed actor guard, BPK/BPX archive boundaries, Saturn font parse/render determinism, save-slot round-trip, mechanics parity slices, and renderer/data loader probes.
Theron V1 Track 02 bank/object marker, save/progress round-trip, shop purchase edges, launcher scan reuse, mechanics/champion/progression probes, and runtime screenshot-readiness manifests.
Cross-cutting Nested ZIP deflate cache materialization, ISO/BIN case-insensitive duplicate-hash scanning, M12 no-data/data-dir cache gates, 19-language launcher and game-domain l10n layout validation, and Tier 1 strict boot coverage.

Detail

DM1 V1

Criterion Score Evidence note
reference_inventory 10/10 DM1 PC 3.4 anchors, ReDMCSB source, Greatstone references, and local evidence manifests remain source-locked.
definition_matrix 10/10 docs/parity/PARITY_MATRIX_DM1_V1.md, DM1 V1 movement/completion evidence, and many pass manifests exist.
launch_smoke 9/10 DM1 V1 launch/menu/title/runtime paths are repeatedly CTest- and probe-gated. The real-media matrix passes direct CLI and the start-menu route for both DOS PC 3.4 ZIP layouts, Atari ST English/German, Amiga and FM Towns; the canonical PC 3.4 ZIP, English Atari ST v1.2 archive, Amiga HD archive and FM Towns route independently verify their complete initial input matrices. French DOS unpacked media retains its separate native route. Archived media is consumed in memory and requires neither an emulator nor a BIOS.
core_input_movement 15/15 Input command queue, keypad route, movement timing/legal-target gates, touch/mouse queue routes, stairs transitions, and cooldown/F0380 ordering are source/runtime-gated.
viewport_ui_render 17/20 Viewport wall/door/ornament/side-content source slices, champion panel/HUD/status boxes, Hall mirror frame/portrait probes, chest UI, and title/swoosh gates are broad; full original pixel-pair coverage is still incomplete.
gameplay_systems 14/15 Combat, creature AI/grouping, inventory/chest, champion lifecycle, spells, projectiles, sensors, save/load, resurrection, food/water, and many regression gates are covered. Remaining point is reserved for broader real-runtime parity depth.
audio_timing 3/10 Audio event mapping and some sound-specific gates exist, including door-bash/no-open coverage; cadence/overlap/original timing proof remains limited.
original_overlay_regression 0/10 Original overlay/capture work has partial artifacts and exact wall matches, but representative movement/HUD/viewport parity is not yet promoted.

DM1 V2

Criterion Score Evidence note
reference_inventory 8/10 DM1 source/original inventory applies; V2 remains presentation over V1 source semantics.
definition_matrix 7/10 V2 completion/status artifacts, phase gates, and CTest-backed source-isolation exist.
launch_smoke 8/10 dm1_v2_runtime_presentation_smoke runs real firestaff --game dm1 paths for V2.0/V2.1 and verifies distinct presented frame receipts.
core_input_movement 8/15 Movement/camera, selected-resolution input mapping, 4K mapping, and phase-gated command adapters are covered.
viewport_ui_render 15/20 HUD overlay, item/render source locks, lighting/effects/VFX, viewport materials, per-mode signatures, and GPU/readback probes exist; finished V2.2 real-art gates remain open.
gameplay_systems 5/15 V2 preserves V1 gameplay ownership through phase gates, but V2-specific gameplay proof is intentionally limited.
audio_timing 0/10 No complete V2 audio/timing parity gate.
original_overlay_regression 0/10 No representative original-overlay regression proof for V2.

CSB V1

Criterion Score Evidence note
reference_inventory 9/10 CSB PC/Atari/Amiga references, source-lock manifests, archive/version inventories, and asset-pair manifests exist.
definition_matrix 10/10 docs/parity/PARITY_MATRIX_CSB_V1.md, completion matrix, and parity surface matrix define the CSB V1 evidence boundary.
launch_smoke 6/10 Hash-matched launch intent, blocked-launch UI, quickplay/load-route source locks, PC real-asset launch boundary, and CSB profile/diagnostic gates exist.
core_input_movement 5/15 Command-chain, input-queue, movement step/rotation, runtime tick, queue overflow, reincarnation, and related gates exist, but end-to-end runtime proof is not complete.
viewport_ui_render 7/20 CSB-specific viewport source slices, CustomBackgrounds, D1/D2/D3 wall/door/ornament lanes, hidden item skip tables, and portrait render handoff are covered; real pixel captures remain open.
gameplay_systems 4/15 Dungeon loader/world, DSA trigger, save runtime boundary, utility/CMP import, Grey Lord/Zokathra/chaos slices exist; broad mechanics parity remains open.
audio_timing 2/10 AMG Utility Disk sound decoding is fixed for documented SND2 files; runtime playback/rate binding and timing proof remain open.
original_overlay_regression 0/10 No CSB V1 representative original overlay regression proof.

CSB V2

Criterion Score Evidence note
reference_inventory 5/10 CSB V2 inherits CSB references but remains presentation-layer evidence.
definition_matrix 4/10 V2 phase-gate and verification-suite tests exist, but no full CSB V2 completion matrix.
launch_smoke 4/10 Presentation-mode and settings probes are wired.
core_input_movement 6/15 Smooth movement, smooth runtime binding, and phase verification are CTest-gated.
viewport_ui_render 10/20 HUD overlay, filter config/chain/dispatch, texture upscale, and V2.2 shape probes are wired and passing.
gameplay_systems 2/15 V2 keeps source gameplay boundaries but does not add broad gameplay parity.
audio_timing 0/10 No CSB V2 audio/timing parity gate.
original_overlay_regression 0/10 No CSB V2 original overlay regression proof.

DM2 V1

Criterion Score Evidence note
reference_inventory 8/10 DM2 PC/French/German/Mac data inventories and SKULL.ASM/source references exist; no ReDMCSB C equivalent.
definition_matrix 4/10 DM2 V1 does not yet have a complete parity matrix, but multiple scoped parity gates now define runtime boundaries.
launch_smoke 4/10 Asset scanner/profile coverage and required-file launch gating exist; full runtime launch proof is still active work.
core_input_movement 3/15 Movement plan/exec chain with door attack, creature push/classification, stair level change, and teleporter resolution implemented; core runtime wiring remains incomplete.
viewport_ui_render 3/20 HUD/minimap/presentation-adjacent gates exist; V1 renderer parity remains open.
gameplay_systems 13/15 Door/button, pressure/trigger, projectile, creature collision/death/drop, weather/timer save, world-state/minimap, shop/combat slices, save-adjacent gates, and skproject combat damage pipeline (CALC_PLAYER_ATTACK_DAMAGE, WOUND_PLAYER, ATTACK_PARTY) are covered.
audio_timing 1/10 Limited timing/weather seed coverage; no broad DM2 audio/timing parity proof.
original_overlay_regression 0/10 No DM2 V1 representative original overlay regression proof.

DM2 V2

Criterion Score Evidence note
reference_inventory 3/10 DM2 V2 inherits DM2 references but remains presentation-layer scaffolding.
definition_matrix 2/10 V2 probes exist, but no full DM2 V2 completion matrix.
launch_smoke 2/10 Some presentation/shape/in-place probes exist.
core_input_movement 1/15 V2 movement evidence is minimal.
viewport_ui_render 7/20 DM2 V2 HUD/in-place/modern assets/shape-related probes exist, but finished rendering proof is sparse.
gameplay_systems 1/15 V2 inherits V1 boundaries; no broad V2 gameplay proof.
audio_timing 0/10 No DM2 V2 audio/timing parity gate.
original_overlay_regression 0/10 No DM2 V2 original overlay regression proof.

DM Nexus V1

Criterion Score Evidence note
reference_inventory 8/10 Saturn DMDF/DGN/S2D/MNS/BPK/BPX inventories and hash/profile references exist.
definition_matrix 4/10 No full Nexus V1 parity matrix, but parser/runtime gate coverage is now substantial.
launch_smoke 4/10 Profile/asset scanner and launcher boundaries exist; real-asset end-to-end handoff remains active.
core_input_movement 2/15 World/save/mechanics slices exist, but movement/input parity is sparse.
viewport_ui_render 10/20 Saturn font parser/render, BPK/BPX archive parsing, DMDF palette/string bounds, DGN malformed actor guard, and renderer probes exist.
gameplay_systems 9/15 Save slot round-trip, mechanics parity, creature/world/dungeon data slices, and loader guards exist.
audio_timing 0/10 Nexus audio/timing parity gates remain open.
original_overlay_regression 0/10 No Nexus representative original overlay regression proof.

DM Nexus V2

Criterion Score Evidence note
reference_inventory 4/10 V2 inherits Nexus references.
definition_matrix 2/10 V2 probe coverage exists, but no full completion matrix.
launch_smoke 2/10 Presentation probes build/run in CI.
core_input_movement 2/15 Touch/smooth movement scaffolds exist, but movement parity is not complete.
viewport_ui_render 9/20 Lighting, atmosphere, upscaler, texture/modern asset, and in-place draw probes exist.
gameplay_systems 1/15 V2 gameplay proof is intentionally limited.
audio_timing 0/10 No Nexus V2 audio/timing parity gate.
original_overlay_regression 0/10 No Nexus V2 original overlay regression proof.

Theron V1

Criterion Score Evidence note
reference_inventory 8/10 Track 02 JP/US provenance, launcher scan reuse, and hash/profile evidence exist.
definition_matrix 7/10 Comprehensive Theron V1 parity definition matrix with 10 categories, 40+ gates, proven/open status, and bound-byte totals.
launch_smoke 7/10 Full IPL-to-stage-2 boot chain verified byte-for-byte: IPL loader, entry path, call graph, dispatch machine, and dynamic payload. Disassembly chain test exercises 17 verification functions.
core_input_movement 13/15 Joypad button-to-action mapping proven for all five groups (I=$01, II=$02, Select=$04, Run=$08, D-pad=$F0) via AND #mask + BEQ/BNE patterns. 64 action sites with combined mask $FF. Command VM 10 handlers classified. Joypad read routine, timer, and IRQ proven.
viewport_ui_render 16/20 All 32 VDC registers proven accessed via st0 #reg. MWR (BAT size), CR (sprite/BG enable), SATB DMA, horizontal timing (HSR/HDR), vertical timing (VPR/VDW/VCR) all proven. VCE palette 438 accesses, VRAM clear/tile transfer proven. Rendering pipeline 5 tiers, 2310 bound bytes.
gameplay_systems 14/15 Save/load progress round-trip, shop purchase edges, dungeon progression, mechanics, champion, object-marker, and Track 02 parser slices are covered.
audio_timing 7/10 CD_PLAY track parameter fully mapped: two code-region sites both load track $0E via LDA #$0E → STA $FF; four data false positives filtered. System card catalog: 279 JSR sites to 17 BIOS vectors.
original_overlay_regression 0/10 No representative original overlay regression proof for Theron.