LearnNewsExamplesServices
Frontmatter
id15443
titleDock tear-out calibration: flagship-density thresholds + matrix platform defaults
stateOpen
labels
enhancementai
assignees[]
createdAtJul 18, 2026, 9:44 AM
updatedAtJul 18, 2026, 9:44 AM
githubUrlhttps://github.com/neomjs/neo/issues/15443
authorneo-fable
commentsCount0
parentIssuenull
subIssues[]
subIssuesCompleted0
subIssuesTotal0
contentTrust
projected
quarantined0
signals[]
blockedBy[]
blocking[]

Dock tear-out calibration: flagship-density thresholds + matrix platform defaults

Open Backlog/active-chunk-7 enhancementai
neo-fable
neo-fable commented on Jul 18, 2026, 9:44 AM

Context

The two receipts-gated ACs retained from #15244 (G1 dock tear-out) at its PR assembly — split per the honest-close-target discipline so the delivered leaf closes truthfully while the empirical calibration work keeps its own gates:

Scope

  1. AC-2 (from #15244): empirical hysteresis calibration on flagship density (20 items / 9 nodes / 6 tab nodes): the inherited 0.8/0.6 detachThreshold/reattachThreshold calibrated for false detaches + re-entry jitter ON the dock surface — never restored to an ancestral number by memory; chosen values carry their calibration receipts. Two binding inputs recorded on #15244's thread: real input only (the #15243 matrix proved activation timings unmeasurable under CDP — synthesized input sustains transient activation indefinitely), and Phoebe's threshold-pair watch (PR #15413's review): the earned-re-entry law makes shallow-exit-immediate-return non-re-entering by design; if real users read that as "sticky," the calibration question moves to the threshold PAIR, not the arming.
  2. AC-6 (from #15244): platform defaults (acquisition/embodiment per OS) selected FROM the #15243 matrix, cited with the receipts — gated on the matrix's macOS column (@neo-fable-clio's #15243, rows 4-7 in @neo-kimi-phoebe's hands).

Sequencing gates (why this is its own leaf)

Behind: #15243's remaining rows/receipts + a real-input session (operator hands or a headed human-driven run). Not implementable from a sandbox alone — the calibration IS the human-input evidence.

Acceptance Criteria

  • Threshold pair calibrated on flagship density with real-input receipts recorded (values may confirm 0.8/0.6 — the RECEIPT is the deliverable).
  • The shallow-exit-return path exercised deliberately; perceived stickiness dispositioned as threshold-pair data (Phoebe's watch).
  • Platform defaults cited from the matrix with per-OS receipts; the DockLayoutAdapter/host wiring consumes them as configs, not constants.

Out of Scope

The tear-out mechanism itself (delivered: contract tier #15407/PR #15408; host tier + e2e via #15244's closing PR); the matrix rows themselves (#15243).

Related

Split-from: #15244 · gates: #15243 (matrix) · inputs: PR #15413 review watch (threshold-pair), #15244 thread (real-input-only receipt).

Decision Record impact: none.

Live latest-open sweep: checked latest 12 open issues at 2026-07-18T07:45Z; no equivalent. A2A in-flight claim sweep: the #15244 lane is mine + Ada's (this split is its assembly act); no competing claim.

Origin Session ID: 89818500-8a12-4162-b41f-8947703b1b06

Retrieval Hint: "dock tear-out threshold calibration flagship density matrix platform defaults real-input"