ワンクリックでManusで任意のスキルを実行

$pwd:

ltx-fidelity-debugging

Name: Ltx Fidelity Debugging
Author: numman-ali

// Use this skill when debugging real LTX-2.3 output quality, running safe manual smoke tests, interpreting stage snapshots, or deciding whether a result is specific enough to count as a clear dog or other visually coherent subject.

Manusで実行

$ git log --oneline --stat

stars:3

forks:0

updated:2026年3月7日 22:48

ファイルエクスプローラー

2 ファイル

SKILL.md

readonly

name	ltx-fidelity-debugging
description	Use this skill when debugging real LTX-2.3 output quality, running safe manual smoke tests, interpreting stage snapshots, or deciding whether a result is specific enough to count as a clear dog or other visually coherent subject.

LTX Fidelity Debugging

Use this skill for the narrow slice where the real LTX-2.3 path runs but the output is still visually wrong, ambiguous, or not strong enough to promote into repo truth.

This skill is for:

real-weight local smoke runs
stage-local visual debugging
safe iteration on the 22B AV path
deciding whether the next suspect is prompt/stage 1, x2/stage 2, or decode/export

Do not use this skill for the common implementation path. The default repo loop still lives in AGENTS.md, MEMORY.md, and scripts/dev.py.

Default workflow

Read docs/research/10-ltx-fidelity-debug-playbook.md.
Use scripts/ltx_debug_smoke.py with a fixed seed and real converted artifacts.
Start with the named profile ladder: safe-smoke, then visual-gate, then coherence.
Keep clean lifecycle on unless you are disproving the lifecycle itself.
Use detached tmux execution for long real-weight smokes.
Inspect stage1, post_x2, and final snapshots before deciding what to fix next.
Review frame_0001, frame_mid, and frame_last before promoting a rung.
Downgrade the claim if the result is ambiguous.

What to load

docs/research/10-ltx-fidelity-debug-playbook.md
docs/research/07-ltx-integration-seams.md
docs/research/09-open-questions-and-validation-plan.md
MEMORY.md

Load these code paths only when you need the implementation seam:

packages/families/ltx/src/ltx/_prompt_encoding_backend.py
packages/families/ltx/src/ltx/_generation_backend.py
scripts/ltx_debug_smoke.py

Rules

Start at the smallest truthful rung, not the highest resolution.
Use real converted artifacts for visual conclusions.
Change one major variable at a time.
Trust stage-local evidence over the final clip when localizing a bug.
Generic quadruped output does not count as dog success.

related-skills.json

同じリポジトリ

swiftui-pro.md

from "numman-ali/mlxr"

Comprehensively reviews SwiftUI code for best practices on modern APIs, maintainability, and performance. Use when reading, writing, or reviewing SwiftUI projects.

2026-03-143

family-bringup.md

from "numman-ali/mlxr"

Use this skill when bringing up a new model family, re-basing an existing family onto the current platform contract, or deciding what belongs in core workflow orchestration versus family or host code.

2026-03-143

validation.md

from "numman-ali/mlxr"

Use this skill when working on MLXR runtime debugging, log triage, validation discipline, benchmark-backed claims, or deciding whether runtime behavior is strong enough to count as evidence rather than a one-off run.

2026-03-143

media-asset-composer.md

from "numman-ali/mlxr"

Design static and light-motion media assets such as posters, slides, cards, flyers, invitations, social graphics, thumbnails, hero images, banners, certificates, README art, and short promo reels. Use when Codex needs to turn text, images, icons, or video into polished layout-driven visuals, especially through a reproducible Python, Pillow, SVG, or ffmpeg workflow.

2026-03-133

package.json

"author": "numman-ali"

"repository": "numman-ali/mlxr"

GitHub リポジトリを開く Creator のリポジトリを見る

$ install --global

$ download --local

Manusで実行

$ useful --forSOC

ソフトウェア開発者コンピュータ・数学職15-1252L4

LTX Fidelity Debugging

Use this skill for the narrow slice where the real LTX-2.3 path runs but the output is still visually wrong, ambiguous, or not strong enough to promote into repo truth.

This skill is for:

real-weight local smoke runs

stage-local visual debugging

safe iteration on the 22B AV path

deciding whether the next suspect is prompt/stage 1, x2/stage 2, or decode/export

Do not use this skill for the common implementation path. The default repo loop still lives in AGENTS.md, MEMORY.md, and scripts/dev.py.

Default workflow

Read docs/research/10-ltx-fidelity-debug-playbook.md.

Use scripts/ltx_debug_smoke.py with a fixed seed and real converted artifacts.

Start with the named profile ladder: safe-smoke, then visual-gate, then coherence.

Keep clean lifecycle on unless you are disproving the lifecycle itself.

Use detached tmux execution for long real-weight smokes.

Inspect stage1, post_x2, and final snapshots before deciding what to fix next.

Review frame_0001, frame_mid, and frame_last before promoting a rung.

Downgrade the claim if the result is ambiguous.

What to load

docs/research/10-ltx-fidelity-debug-playbook.md

docs/research/07-ltx-integration-seams.md

docs/research/09-open-questions-and-validation-plan.md

MEMORY.md

Load these code paths only when you need the implementation seam:

packages/families/ltx/src/ltx/_prompt_encoding_backend.py

packages/families/ltx/src/ltx/_generation_backend.py

scripts/ltx_debug_smoke.py

Rules

Start at the smallest truthful rung, not the highest resolution.

Use real converted artifacts for visual conclusions.

Change one major variable at a time.

Trust stage-local evidence over the final clip when localizing a bug.

Generic quadruped output does not count as dog success.

ltx-fidelity-debugging

LTX Fidelity Debugging

Default workflow

What to load

Rules

このリポジトリの他の Skills

このリポジトリの他の Skills

LTX Fidelity Debugging

Default workflow

What to load

Rules