Repository navigation
mihara-bonding-chain: live Restraint pacing + copyResource burst stack-mirror - #147
Merged
Merged
Conversation
…ver 4s Owner measurement: Restraint Chains are consumed at 0.4s per stack, delivering Ensnaring stacks gradually rather than instantly at battle start / fullBurstEnd. - Replace the two instant +10 resource deltas with ten +1 deltas at 0.4s intervals (delaySec 0, 0.4, ..., 3.6). - Unit tests pass (15/15). - Board ratio: 1.179 -> 1.177 (N6, n=2). Channel split total: 895.2M -> 891.3M (-0.4%). This is a real fidelity improvement but does NOT carry the 17.9% over-model; the residual remains after coefficient, stack-count, core, and buff-scope checks.
…acks over 4s Owner measurement: Restraint Chains are consumed at 0.4s per stack. Model the full 10-chain dump as ten separate 50.06% flatDamage hits and ten separate +1 Ensnaring resource deltas, both staggered by 0.4s. - Replace two instant 500.6% damage blocks with twenty paced 50.06% blocks (10 at battle start + 10 per fullBurstEnd). - Replace two instant +10 Ensnaring resource blocks with twenty paced +1 blocks on the same timing. - Update unit spec to assert ten paced hits per dump and the single-chain counterfactual (keep one hit per trigger). - Update mihara-channel-split.ts to classify 50.06% hits as Restraint. Board ratio: 1.179 -> 1.174 (N6, n=2). Total: 895.2M -> 891.0M. Unit tests pass; typecheck passes.
…onsumed during FB Owner measurement: Restraint Chains are consumed at 0.4s per stack. Key insight: the S2 +1-per-40-normals-in-FB trigger is the same event as consuming one Restraint Chain (deals 50.06% and adds +1 Ensnaring). The 10-chain recharge at fullBurstEnd is then consumed over 4s after each FB end. - Add resource [0..10], initial 10, recharged to 10 at every fullBurstEnd. - Replace S1 dumps with ten paced consumption blocks at fullBurstEnd: -1 restraint, +1 ensnaring, 50.06% flatDamage, delaySec 0..3.6. - Replace S2 +1 ensnaring block with: -1 restraint, +1 ensnaring, 50.06% flatDamage, gated to in-FB and restraint>=1. - Update unit spec: dumps no longer at battle start; cap assertion allows fixture-specific 19-stack peak. - Update caveats: note the fullBurstEnd consumption fires every FB end because measured stack behavior and board ratio require it. Board ratio: 1.179 -> 1.020 (N6, n=2). Total: 895.2M -> 777.4M. Unit tests pass; typecheck passes.
- Treat Restraint Chains as a live resource. S2's +1 Ensnaring per 40 normals in FB now consumes one restraint and adds one ensnaring in the same event. At every fullBurstEnd the resource recharges to 10 and is then consumed over 4s via ten 0.4s-paced blocks (50.06% each + +1 ensnaring). This fires on every FB end, matching the measured 20-stack Ensnaring cap and the board-ratio correction (N6 1.179 -> 1.020). - Update mihara-bonding-chain unit tests (15/15) and regression snapshot. - Refresh kit-status, unmodeled-review, STATE.md, engine-modeling-gaps. - Update block-order census fixture for the new restraint producer/consumer pairs (15 units, 145 pairs). Verified: bash scripts/verify.sh green (319 test files, 4794 passed). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- Gate the S1 Restraint Chain recharge/dump to the end of Mihara's own Full
Burst only (ownBurstGate:'cast'). It no longer fires on every team FB end.
- Remove the Restraint consumption and 50.06% flatDamage from S2's
40-normals-in-Full-Burst trigger; it now simply adds +1 Ensnaring stack.
This matches the kit line and the intended cycle:
mihara FB end -> +10 Ensnaring from restraint dump
other FB -> +~10 Ensnaring from S2 procs
mihara next burst -> consumes 20 Ensnaring into Dragging Chain.
- Update unit-test fixture to the two-B3 comp and assert dump clusters are
anchored to mihara's own FB-end windows.
- Refresh regression snapshot, block-order fixture (144 pairs), kit-status,
unmodeled-review, engine-modeling-gaps, dpschart, and rank boards.
Verified: bash scripts/verify.sh green (319 test files, 4794 passed).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…hain burst DoT with live Ensnaring stacks - Add `copyResource` effect kind: copies one owner resource pool into another, clamped to the target's [min,max]. - Use it in mihara-bonding-chain burst: snapshot `ensnaring` into `mirror`, cancel `ensnaring`, then tick `mirror × 50.05` for 10s. - Update unit tests to assert live stack scaling, low first burst, and later bursts reaching 1001%/s. - Refresh kit-status, unmodeled review, regression snapshot, dpschart/rank boards. Verified: bash scripts/verify.sh green. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Conflict resolution: mihara override takes the branch's live-restraint model (main's side was note prose describing the superseded model); validator keeps both new effect checks (copyResource + selfStatus); kit-status/gaps mirrors regenerated via --sync-mirrors + doc-drift --update; regression snapshot regenerated on the merged tree — footprint vs origin/main is the N6 mihara comp only (mihara-bonding-chain −9.31%, little-mermaid +0.07%). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…anding DECISIONS 2026-08-25 entry (live Restraint pacing + copyResource burst mirror, measured −9.31% on the N6 graded comp matching the 2026-08-17 localization, board 1.179→1.076 HOT); STATE.md copyResource effect row; fit-exposure handoff CLOSED and archived (localization artifact's priorHandoff repointed); 2026-08-24 followups stack-mirror bullet updated to landed-state; override note carries the fresh board figure + 16-test spec citation; mirrors re-synced. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…eader re-derivation, resourceGate caveat, dead line + probe label Cross-family review (kimi-code/k3) returned FIX-BEFORE-MERGE; both FIX findings were stale/incorrect PROSE describing the shipped FB-end-anchored pacing — code unchanged. Census-scope + unmodeled-review-heuristic follow-ups filed in the 2026-08-24 followups handoff. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…BurstGate check against vacuous pass; file validator declared-resources follow-up Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Re-models
mihara-bonding-chain's two-currency stack kit as live resources (session rulings 2026-08-21):restraint[0..10]: initial 10, refilled to 10 at the end of HER OWN Full Burst only (ownBurstGate:'cast'), consumed one stack per 0.4s over the 4s after that FB end (the measured "specific timing" anchor) — each consumption deals one 50.06% hit and feeds +1 Ensnaring. No battle-start dump.mirrorpool at cast via the newcopyResourceengine effect (clamp-copy of one named pool into another at apply time; copy lands before the same-cast pool zeroing), then cancels Ensnaring; the DoT ticks mirror × 50.05%/s for 10s — 1001%/s only when the pool actually reached 20.Closes the 2026-08-17 "LOCALIZED v2" over-model QUEUE item (burst DoT assumed a 20-stack pool) and enacts the previously-gated stack-mirror cast-snapshot primitive with
mihara-bonding-chainas its first consumer (asuka-wille's finisher 30-cap stays gated on its own measurement).Measured effect
Snapshot footprint vs origin/main is exactly the "N6 mihara/maiden wind" comp:
mihara-bonding-chain−9.31% (the 2026-08-17 localization predicted ~9.3%),little-mermaid+0.07%. Board 1.179 → 1.076 HOT (comps 1.03 / 1.12). Full-burst counts unchanged (regression N6 asserts green).Gates
verify.shfully green on the merged tree (321 test files; board artifacts rebuilt).scripts/tests/units/mihara-bonding-chain.test.ts: 16 tests (M1–M5) green, incl. the exact 20-cap pin (501.6) and a vacuous-gate guard on the ownBurstGate check.docs/handoffs/2026-08-24-kit-audit-primitive-followups.md.copyResourcerow; fit-exposure handoff closed.🤖 Generated with Claude Code