Навыки video-inpainting

📦

video-inpainting

Name: video-inpainting
Author: prime-skills

Ревизия содержимого r1 Средний риск ⚙️ Внешние команды🌐 Доступ к сети📁 Доступ к файловой системе

Edit Video Regions with RunComfy

Unwanted objects and inconsistent regions can distract viewers across an entire video. This skill guides RunComfy model selection and regional editing prompts while identifying tasks that require separate masking workflows.

Поддерживает: Claude Codex Code(CC)

📊 70 Адекватно

Maintain this Skill? Claim it 📊 70 Адекватно

Установить с помощью моего Агента

Скопируйте этот запрос в своего Агента. Он содержит каноническую страницу Skill и манифест.

Запрос агента

Review the Skillstore skill "video-inpainting" from https://skillstore.io/skills/prime-skills-video-inpainting.md and its manifest at https://skillstore.io/api/skills/prime-skills-video-inpainting/manifest. Verify the artifact. You may proceed after verification, subject to the environment's own policy.

Ваш Агент по-прежнему должен показать план и запросить все подтверждения, требуемые политикой безопасности.

Ресурсы для AI-агентов

Используйте эти ссылки, когда AI-агенту, crawler или script нужен чистый контекст вместо полной страницы.

Markdown-детали GET /skills/prime-skills-video-inpainting.md Подписанный manifest GET /api/skills/prime-skills-video-inpainting/manifest Подписанный lockfile GET /api/skills/prime-skills-video-inpainting/lockfile

Протестировать

Использование «video-inpainting». Recommend a way to remove an overhead cable from my clip.

Ожидаемый результат:

Recommended route: Wan 2-7 Edit-Video for a prompt-driven regional cleanup.
Suggested prompt: Remove the overhead cable across all frames and reconstruct the matching background. Preserve all other content.
Review camera movement and repaired edges before accepting the result.

Использование «video-inpainting». I need exact masks around a moving object across the whole clip.

Ожидаемый результат:

Use a separate ComfyUI workflow with segmentation tracking and temporal inpainting.
The documented CLI endpoints cannot provide the required pixel-precise mask propagation.
Review the LTX 2-3 inpainting workflow in the RunComfy GUI before processing the full clip.

Использование «video-inpainting». Can I edit every frame independently and rebuild a long video?

Ожидаемый результат:

Seedream sequential editing is suggested for short frame sequences.
Independent still edits can lose temporal consistency on long clips.
Frame extraction and video re-encoding require separate tools.

Аудит безопасности

Средний риск

v1 • 01.11.2026 Открыть версионный отчет

All 57 static findings are false positives involving Markdown, reference URLs, standard authentication documentation, or model-selection advice. One semantic concern remains: the guide incorrectly treats JSON input as protection against shell injection during command construction. No evidence found of malicious instructions, credential exfiltration, or security-review manipulation; external CLI behavior was not verified.

Просканировано файлов

176

Проанализировано строк

Пункты проверки

Ложные срабатывания проигнорированы

Подтверждённые проблемы безопасности (1)

Средний

SKILL.md:77-84 SKILL.md:161

Shell Safety Claim Omits Caller-Side Quoting

The guide claims "No shell-injection surface" while placing user prompts and URLs inside shell-single-quoted JSON. An unescaped apostrophe can end that quoting before the CLI receives input; JSON does not protect the invoking shell.

The example and absolute safety claim are visible, but no CLI implementation or dynamically assembled command is included. Exploitability depends on how an agent inserts user input.

Факторы риска

⚙️ Внешние команды (31)

🌐 Доступ к сети (21)

📁 Доступ к файловой системе (2)

SKILL.md:160 SKILL.md:160

Аудитор:: codex

Поделиться и цитировать этот отчет

Делитесь версионным отчетом об оценке, нейтральным значком, встраиваемой карточкой и цитатами. Skillstore публикует доказательства, не решая, безопасен ли этот Skill.

Открыть версионный отчет

Копировать ссылку на отчёт

https://skillstore.io/skills/prime-skills-video-inpainting/audits/1?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_report

Значок Markdown

[![Skillstore security assessment](https://skillstore.io/badges/skills/prime-skills-video-inpainting/security.svg)](https://skillstore.io/skills/prime-skills-video-inpainting?utm_source=security_passport_badge)

Значок HTML

<a href="https://skillstore.io/skills/prime-skills-video-inpainting?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/prime-skills-video-inpainting/security.svg" alt="Skillstore security assessment" loading="lazy"></a>

Встраиваемая карточка

<iframe src="https://skillstore.io/embed/skills/prime-skills-video-inpainting.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>

Академические ссылки (APA · BibTeX · CFF)

Цитата APA

prime-skills. (2026). video-inpainting security audit report (audit version 1) [Author version unspecified]. Skillstore. https://skillstore.io/skills/prime-skills-video-inpainting/audits/1

Цитата BibTeX

@techreport{prime-skills-prime-skills-video-inpainting-2026,
  author = {prime-skills},
  title = {video-inpainting security audit report (audit version 1)},
  institution = {Skillstore},
  year = {2026},
  number = {1},
  url = {https://skillstore.io/skills/prime-skills-video-inpainting/audits/1},
  note = {Author version unspecified}
}

CITATION.cff

cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "video-inpainting security audit report (audit version 1)"
version: "unspecified"
type: report
authors:
  - name: "prime-skills"
date-released: "2026-11-01"
url: "https://skillstore.io/skills/prime-skills-video-inpainting/audits/1"
identifiers:
  - type: other
    value: "skillstore:prime-skills-video-inpainting:audit:1"
    description: "Skillstore immutable audit report identifier"

Сравнить варианты

4 устанавливаемых вариантов

Навык каждого автора остаётся отдельным устанавливаемым вариантом. Рекомендуемый вариант ранжируется по данным Skillstore.

Почему этот вариант первый

Самая высокая оценка Skillstore

Автор / источник Оценка Skillstore Достоверность доказательств Использование в Skillstore Обновлено

agentspace-so Рекомендуется

agentspace-so-video-inpainting

Оценка Skillstore 78

Достоверность доказательств Высокий

Использование в Skillstore 36

Обновлено

2026-10-22

runcomfy-com

runcomfy-com-video-inpainting

Оценка Skillstore 71

Достоверность доказательств Высокий

Использование в Skillstore 47

Обновлено

2026-09-09

doany-ai

doany-ai-video-inpainting

Оценка Skillstore 70

Достоверность доказательств Высокий

Использование в Skillstore 45

Обновлено

2026-09-09

prime-skills Текущий

prime-skills-video-inpainting

Оценка Skillstore 70

Достоверность доказательств Средний

Использование в Skillstore Неизвестно

Обновлено

2026-11-01

Оценка Skillstore

Почему такая оценка Достоверность доказательств: Средний

Архитектура

Сопровождаемость

Контент

Сообщество

Соответствие спецификации

Что вы можете построить

Clean Up Background Distractions

Plan a regional edit to remove overhead cables or a passing background object while preserving the surrounding scene.

Plan Consistent Product Replacements

Compare prompt-driven replacement, identity-aware restyling, and separate frame workflows for an authorized product video update.

Select a Masking Workflow

Determine when a scientific or training clip needs precise masks and temporal propagation instead of a prompt-only model call.

Попробуйте эти промпты

Choose a Basic Cleanup Route

I own this clip: [video URL]. Recommend a route to remove the overhead cable while preserving everything else. Do not submit a job yet.

Prepare a Regional Edit

Prepare a Wan edit for [video URL] to remove the box in the lower-left corner. Preserve surrounding content. Show the prompt and execution plan before submission.

Compare Identity-Sensitive Routes

Compare Wan and Lucy for replacing the jacket in [video URL] while preserving the subject's identity. Identify limitations and recommend a small test before submission.

Design a Precise Multi-Pass Workflow

Plan separate edits for [video URL]: remove a moving object, then replace a sign. Compare prompt-only editing with tracked ComfyUI masks. Include temporal checks and shell-safe input handling.

Лучшие практики

Use source URLs explicitly provided for the task, and confirm authorization before sharing footage with an external service.
Describe one spatially specific change per pass, request preservation, and inspect motion continuity after each edit.
Serialize inputs and use argument-array execution where available; do not interpolate prompts or URLs directly into shell commands.

Избегать

Combining unrelated regional edits into one prompt and expecting unchanged surrounding frames.
Treating prompt-driven endpoints as substitutes for precise masks or guaranteed temporal propagation.
Assuming JSON prevents shell injection or treating unverified CLI privacy claims as established guarantees.

Часто задаваемые вопросы

What must I install before using this skill?

The guide requires the RunComfy CLI and a signed-in account. Network access and access to the selected model are also required.

Which model is the default?

Wan 2-7 Edit-Video is the default for prompt-driven regional edits. Lucy and Seedream are alternatives for specific editing requirements.

Can the CLI propagate exact masks?

Not through the documented endpoints. The guide directs precise masking and temporal propagation tasks to separate ComfyUI GUI workflows.

Will all other content remain unchanged?

No guarantee is provided. Explicit preservation prompts and one change per pass can reduce drift, but every result requires visual review.

Does this skill process video locally?

No local video-processing engine is included. The documented CLI sends requests to RunComfy, polls job status, and downloads results.

Is JSON input sufficient protection against shell injection?

No. JSON structures the model request but does not escape the invoking shell. Use structured serialization and safe argument passing for user input.

Сведения для разработчиков

Автор

prime-skills

Лицензия

MIT

Ревизия Skillstore

Примечание о версии

Автор не указал версию.

Репозиторий

https://github.com/prime-skills/runcomfy-agent-skills/tree/fca19ae084c2ae4667f5c8affbb99da5103d5832/video-inpainting

Ссылка

a53bacaa2e85e44d9f929fa7a6d863cbe7fd1c03

Актуальность поддержки

01.11.2026

Использование

0 загрузок · 0 просмотров

Структура файлов

📄 SKILL.md