Changelog

1.0.4 - Input window and failure observability

• Accept input timestamps symmetrically and grow the window by half of the measured per-client round trip, capped at 150ms. A client whose estimated server clock runs faster than the host, or whose press arrives later than the previous 180ms lag limit, no longer loses that press silently. A timestamp outside the window is still ignored rather than charged as a failure, so one slow packet cannot cost the attempt its 25% jam roll. The arc test itself is unchanged.
• Report rejected protocol packets, out-of-window timestamps and unavailable round-trip measurements as rate-limited warnings. A host/client version or session mismatch is now visible in the log instead of leaving the hover tip on "Syncing safe..." forever.
• Undo a partially applied patch set when startup fails, and degrade instead of throwing when a Harmony injection point is missing. A missing movement-input or scan-colour injection now disables that single feature; safes, rewards and audio keep working.
• Remove the decoder's consumption weight exactly once. The previous unconditional subtraction ran a second time when another mod already repairs the pocketed-item path (for example DestroyItemInSlotFix), leaving the operator lighter than the inventory states.
• Show a quiet local cue when a press is never acknowledged, instead of letting the operator discover it from the step timeout.
• Add offline coverage for the symmetric window, round-trip compensation, rejection logging and decoder weight bookkeeping, and compare the plugin metadata against the assembly version so a stale DLL cannot be verified as the current build.
• Not verified on a real second machine, with more than two players, with reconnects or late joins, or with human input. The two-process localhost LAN acceptance recorded for 1.0.3 has not been repeated for this build. Gameplay defaults and existing configuration are unchanged.

1.0.3 - Remote interaction repair

• Validate named-message payload sizes relative to the reader's current position. Unity Netcode has already consumed its named-message hash when it invokes the handler; treating the full reader length as the payload silently rejected remote handshakes, opening requests and safe snapshots.
• Preserve strict message boundaries, session identity and host authority.
• Add nonzero-reader-offset regression coverage and verify the actual v81 NGO named-message framing contract inside the game.
• Verify remote native interaction, QTE completion, decoder pickup/installation/ consumption, and both reward batches with two real localhost LAN game processes, including an isolated copy of the current development modpack. Both peers resolved matching reward IDs and initial values without duplicate applications.
• Require GoldScrapLite 1.0.3. Every player must update both mods together.
• Cross-computer Steam play, larger groups, reconnects and late joins are not covered by this local automated test. Gameplay defaults and existing configuration are unchanged.

1.0.2

• Require GoldScrapLite 1.0.2 so new installations receive the default-on optional safe-only gold spawning setting. It can be disabled in GoldScrapLite to retain natural gold spawns alongside safe rewards.

• Prefer safe placement with the back against a nearby solid wall and the door facing usable open floor. Retain reachable-floor, body, open-door and walking-route clearance checks, and fall back to clear indoor floor when no valid wall position is available.

• Preserve the guaranteed deep-interior safe, randomly distributed extras and default 12-metre separation. No new configuration settings are required.

• Keep spawn counts, gold rewards, QTE behavior and audio unchanged.

1.0.1

• Replace the package cover with the approved AI-generated artwork: the industrial safe slightly ajar, leaking golden light inside a facility surrounded by creatures.
• Update cover credits and AI disclosure in English.
• Packaging and artwork update only. The game DLL, asset bundles, default configuration and dependencies are identical to 1.0.0. The in-game plugin version remains 1.0.0.

1.0.0

• First stable release, combining the gold-safe timing challenge, single-use decoder, mechanical audio and native green scan markers.
• Reserve one guaranteed safe plus independent extra slots by native Risk Level: D: one 10% slot; C/B: two 30% slots; A/S/S+/S++: ten 50% slots.
• Prefer deep interior floor for the first safe; randomly distribute extras across the whole facility with a configurable 12-metre default minimum spacing.
• Search expanded room and connected interior NavMesh samples, with a bounded first-safe fallback and up to three attempts without rerolling the round quantity.
• Preserve floor, body, doorway and walking-route checks, and exclude non-facility destinations. Placement still requires usable indoor floor; optional extras are skipped when there is insufficient safe-sized space or separation.
• Keep both testing spawn options disabled by default.

0.4.4

• Use the native green scan appearance for industrial safes and loose decoders.
• Keep both scan nodes in the general-object gameplay category and preserve the decoder's utility-item type, without changing scrap counts, values or saves.
• Apply the color only when the native scanner animates this mod's markers; other objects keep their normal scanner colors.

0.4.3

• Add native right-click scan markers for industrial safes and loose decoders, with a 20-metre range and normal line-of-sight checks.
• Show the safe's current opening state without revealing its hidden reward plan.
• Hide a decoder's loose-item scan marker when it is installed.
• Preserve decoder utility-item classification, save identity, configuration, audio, reward probabilities and timing.

0.4.2

• Add a spatial, mechanical combination-dial loop while an installed decoder works.
• Fix the missing success cue on the final QTE hit. Both manual and automatic completion now play the original success recording followed by the original opening recording with a short 0.15-second lead.
• Stop the decoder loop at completion, view retirement and session cleanup; do not replay old opening sounds for late joiners.
• Preserve immediate QTE dismissal, existing approved cues, gameplay values, test switches, original save data and other mod configuration.

0.4.1

• Add TestSpawnDecoderBesideSafe, a host-only test switch that places one extra decoder in front of each successfully spawned safe. Default: false.
• Keep normal 5% bonus drops, 15-second decoding, gold rewards and existing gameplay settings unchanged. Test placement uses clear floor in front of the safe and does not retire a valid safe when an item fails to spawn.

0.4.0

• Add the supplied DIAL-LINK 02 decoder as one-handed, ship-saveable equipment.
• Independently roll a 5% extra decoder drop for each successfully opened safe, separately from the original 1–5 gold reward count and item selection.
• Attach a decoder to a closed or jammed safe to open it automatically after 15 seconds. It is spent on installation and remains visually on the door.
• Synchronize installation, remaining time and opening from the host, including late joiners. Leaving, dying or disconnecting does not stop an installed unit.
• Add host settings for decoder drop chance and opening delay. Keep existing gold, QTE, audio, placement, other-mod configuration and save data intact.

0.3.0

• Preselect and retain each safe's gold count when it spawns. By default one to five gold items require three to seven successful QTE steps. Opening pays out the retained count; failure, cancellation, changing operators and late joining cannot reroll it. Keep the 70/20/6/3/1 reward weights.
• Keep the approved difficulty progression by step number and cap it at step five. Steps five through seven share the same 1.25-second period and 10% target width; shorter safes do not compress the progression to reach the cap early.
• Show one centered mark per completed successful press, with no empty slots or advance display of the safe's total step count or gold count.
• Add RewardLinkedQte and ExtraQteSteps; retain RequiredSteps as the fixed mode when linking is disabled. Keep recorded sounds C/B/F and private protocol 3; the existing per-safe required-step snapshot carries the different lengths.
• Preserve the independent runtime recovery, one-time reward delivery, late-join synchronization and session-end cleanup. Live facility and multiplayer acceptance remain separate from offline and isolated startup checks.

0.2.3

• Use approved audition C (the metal-opening recording) for each confirmed QTE success; retain audition B for jamming, F for opening, and all existing settings.

0.2.2

• Use the approved 256 × 288 pixel dial at the same display size. Keep point filtering and the transparent background, halve the visible black-outline thickness, and remove the extra drop shadow. Timing and hit angles are unchanged.
• Replace confirmed-progress and jam feedback with the approved CC0 lock and metal-closure recordings. Use the approved squeaky metal hinge recording for opening. Embed the exact audition files; retain the existing start, press and ordinary-failure cues and local volume control.
• Add recording sources, authors, CC0 notices and processing details in AUDIO_CREDITS.txt; retain the Codex assistance disclosure.
• Set default reward-count weights to 70%, 20%, 6%, 3% and 1% for one through five gold items. Existing configuration values remain in place when replacing the DLL.
• Retain operator movement/look locking, immediate dial dismissal on final success, session-end cleanup, gameplay timing, 25% jam chance and private protocol 3.

0.2.1

• Add black pixel outlines and a shadow to the transparent dial so its indicator, target and prompt remain distinct against flashlight-lit surfaces.
• Lock only the active local operator's movement and camera input during the challenge. Keep Space and Esc available; release the lock when the attempt finishes, fails or is cancelled, without disabling the whole player update.
• Add short, procedurally generated mechanical feedback and a local AudioVolume setting (0–1, default 0.6). No external sound assets are bundled.
• Hide the dial immediately after the final successful press, including while host confirmation is pending. Do not leave it rotating or display a lingering Open state; retain the door animation and opening sound.
• Clear temporary safe objects, active attempts and pending safe-reward timers when the session ends. This addresses an opened safe surviving a return to the main menu and then appearing opened or stuck at "Syncing safe..." after rehosting the same save. The live quit/rehost sequence still needs verification.
• Change safe rewards from 3–5 to a weighted 1–5 count: 60%, 20%, 10%, 8% and 2% respectively. Add RewardCountWeights; minimum/maximum bounds filter the count outcomes and normalize their weights. Eligible gold item types remain equally weighted, independently of the reward-count distribution.
• Keep the five-step difficulty, 30ms edge tolerance, failure probability, placement, gold item effects and private protocol 3 unchanged.

0.2.0

• Replace the timing bar with the approved circular floating-indicator QTE: transparent background, orange pixel ring, white indicator, Space prompt and five progress marks. Remove the panel, large heading and permanent rule text.
• Keep clockwise rotation continuous through successful steps. Prepare once at the start of an attempt; move each new target ahead with at least 350ms of lead.
• Increase difficulty evenly from 1.85s per revolution / 18% arc to 1.25s / 10%. Add configurable final speed, final arc width and 30ms tolerance per arc edge.
• Capture the original press time and grade its recently displayed frame. Keep input and hit angles continuous; pixelation affects only the rendered texture.
• Upgrade the private protocol to version 3 for continuous-phase timing snapshots. All players need the same updated mod. Jam chance, placement and gold rewards retain their existing rules.

0.1.2

• Initialize HDRP Lit lighting stencil, passes and texture keywords using the game's material validator; preserve ordinary lighting/shadows, without emission.
• Use the native hand interaction sprite instead of a null sprite that becomes a square at the center of the screen.
• Replace the three-attempt cap with an independent configurable 25% jam chance on each failure. The host determines and synchronizes each result exactly once.
• Upgrade the private network protocol to version 2; all players need this version.
• Keep the existing timing UI layout pending approval of a circular replacement.

0.1.1

• Keep runtime logic and patches alive across the pack's early bootstrap cleanup; restore the independent driver after scene transitions.
• Add TestSpawnNearEntrance: place one safe on reachable clear floor near the interior main entrance, ignoring normal chance/count/depth for convenient testing.
• Log generation scheduling, runtime recovery and entrance placement outcomes.

0.1.0

• First local experimental package: industrial safe, indoor placement, shared timing challenge and GoldScrapLite rewards.
• Configurable placement, attempts, timing, reward counts/value and exclusions.
• Authoritative attempt state and separate reward-value synchronization.