ワンクリックでManusで任意のスキルを実行

perceptual-listening-harness

スター2

フォーク0

更新日2026年3月1日 18:54

Design and execute blinded perceptual listening studies for LocusQ (trial schema, randomization, metrics, statistical gating, reproducibility artifacts) aligned with BL-060 and BL-061 promotion decisions.

インストール

Codex または Claude でインストールこの Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。

Manusで実行

ソース

joshband

joshband/LocusQ

GitHub リポジトリを開く Creator のリポジトリを見る

ダウンロード

Manusで実行

Perceptual Listening Harness

Use this skill for controlled listening-test design and evidence-driven promotion decisions.

Scope

Blind protocol design (2x2 and related variants).
Trial schema and artifact contracts (CSV/TSV/summary markdown).
Objective metric extraction (MAE, front/back confusion, externalization summary).
Statistical gate decisions for backlog promotion.

Workflow

Freeze protocol before collecting data.
- Define participant count, scene count, condition matrix, and randomization policy.
Enforce blind run discipline.
- Hide condition labels during response capture.
Capture machine-readable trial outputs.
- Include true angle, response angle, error, condition, and reaction time.
Compute and publish gate metrics.
- Per-condition MAE/confusion/externalization.
- Statistical significance or predefined threshold outcome.
Publish reproducibility packet.
- Re-run analysis and ensure stable summary outputs.

Required Evidence

trial_log.csv
metrics_summary.tsv
stats_report.md
gate_decision.md
reproducibility_check.tsv

Cross-Skill Routing

Pair with hrtf-rendering-validation-lab for offline/realtime render parity baselines.
Pair with spatial-audio-engineering for coordinate/scene semantics and renderer assumptions.
Pair with skill_docs for runbook/index promotion wording and evidence linkage.

References

references/trial-schema.md
references/gate-metrics.md
references/reproducibility-contract.md

Deliverables

Protocol summary and rationale.
Explicit gate decision with evidence paths.
Validation status: tested, partially tested, or not tested.

このリポジトリの他の Skills

同じリポジトリ

skill-docs

joshband/LocusQ

Documentation governance skill for metadata compliance, ADR hygiene, invariant traceability, and validation evidence logging.

2026-03-182

documentation-hygiene-expert

joshband/LocusQ

SDLC-aware documentation cleanup and governance specialist for de-bloating repos, consolidating source-of-truth docs, enforcing freshness ownership, aligning implementation docs with ADRs, and automating git artifact hygiene (tracked ignored files, stale archives, and history bloat audits).

2026-03-182

apple-spatial-companion-platform

joshband/LocusQ

Swift companion platform and Apple API ownership for LocusQ. Use when user requests AirPods workflows, CoreMotion motion capture, ear-photo/depth capture, privacy-retention policy, or Apple API boundary decisions for BL-057/BL-058.

2026-03-182

auv3-plugin-lifecycle

joshband/LocusQ

End-to-end AUv3 support workflow for LocusQ. Use when user requests AUv3 format enablement, app-extension architecture, JUCE integration boundaries, AUv3 host validation lanes, or cross-format parity decisions.

2026-03-182

documentation-hygiene-expert

joshband/LocusQ

SDLC-aware documentation cleanup and git artifact hygiene for LocusQ. Use when user requests de-bloating docs, consolidating source-of-truth files, fixing stale ADR alignment, backlog/runbook clarity passes, or git tree artifact cleanup.

2026-03-182

headtracking-companion-runtime

joshband/LocusQ

Companion runtime state-machine and diagnostics for LocusQ head-tracking. Use when user requests readiness/sync gating, axis/frame sanity checks, pose freshness validation, or deterministic operator diagnostics for BL-053/BL-058.

2026-03-182

name	perceptual-listening-harness
description	Design and execute blinded perceptual listening studies for LocusQ (trial schema, randomization, metrics, statistical gating, reproducibility artifacts) aligned with BL-060 and BL-061 promotion decisions.

Title: Perceptual Listening Harness Skill Document Type: Skill Author: APC Codex Created Date: 2026-03-01 Last Modified Date: 2026-03-01

Perceptual Listening Harness

Use this skill for controlled listening-test design and evidence-driven promotion decisions.

Scope

Blind protocol design (2x2 and related variants).
Trial schema and artifact contracts (CSV/TSV/summary markdown).
Objective metric extraction (MAE, front/back confusion, externalization summary).
Statistical gate decisions for backlog promotion.

Workflow

Freeze protocol before collecting data.
- Define participant count, scene count, condition matrix, and randomization policy.
Enforce blind run discipline.
- Hide condition labels during response capture.
Capture machine-readable trial outputs.
- Include true angle, response angle, error, condition, and reaction time.
Compute and publish gate metrics.
- Per-condition MAE/confusion/externalization.
- Statistical significance or predefined threshold outcome.
Publish reproducibility packet.
- Re-run analysis and ensure stable summary outputs.

Required Evidence

trial_log.csv
metrics_summary.tsv
stats_report.md
gate_decision.md
reproducibility_check.tsv

Cross-Skill Routing

Pair with hrtf-rendering-validation-lab for offline/realtime render parity baselines.
Pair with spatial-audio-engineering for coordinate/scene semantics and renderer assumptions.
Pair with skill_docs for runbook/index promotion wording and evidence linkage.

References

references/trial-schema.md
references/gate-metrics.md
references/reproducibility-contract.md

Deliverables

Protocol summary and rationale.
Explicit gate decision with evidence paths.
Validation status: tested, partially tested, or not tested.