video-reference

video-reference is a skill for Claude Code from zenstory-ai/video-recap-skills. It costs 98 tokens per session (2,936 once invoked), scanned A, a copy of video-reference, MIT.

A tool for analyzing a finished video and turning its production methods and measurements into a reusable production-reference file. It can examine cuts, shot timing, loudness, sections, and audio-track roles while keeping source-specific facts separate from general methods.

In plain words
What is it for?
Use it to measure an MP4, MOV, MKV, or WebM, inspect possible edit points and long shots, validate the measurements, and export production_reference.json for another project.
Why use it?
It gives future productions concrete reference data without putting the original video's names or dialogue into the exported file. It is an on-demand analysis step, not a production or quality-control pipeline.

Skill for Claude Code

Written for Claude Code: user-invocable in frontmatter.

Part of the video-recap-skills plugin — 7 skills shipped together

Good fit Use it to measure an MP4, MOV, MKV, or WebM, inspect possible edit points and long shots, validate the measurements, and export production_reference.json for another project.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/zenstory-ai/video-recap-skills/video-reference
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent
npx skills add zenstory-ai/video-recap-skills --skill video-reference
Clone the repo
git clone --depth 1 https://github.com/zenstory-ai/video-recap-skills

Made for: Claude Code.

Or install video-recap-skills, the plugin that ships this one along with the rest of its 7 skills.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for video-reference

README.md
[![agentmods](https://agentmods.dev/badge/skills/zenstory-ai/video-recap-skills/video-reference/github.svg)](https://agentmods.dev/skills/zenstory-ai/video-recap-skills/video-reference)
Your own site
<a href="https://agentmods.dev/skills/zenstory-ai/video-recap-skills/video-reference"><img src="https://agentmods.dev/badge/skills/zenstory-ai/video-recap-skills/video-reference/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for video-reference

Your own site · 80×15
<a href="https://agentmods.dev/skills/zenstory-ai/video-recap-skills/video-reference"><img src="https://agentmods.dev/badge/skills/zenstory-ai/video-recap-skills/video-reference.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 98 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 2,936 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin 100% copy Near-identical to another mod in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00098 $0.02936
Opus 5.5 $0.00039 $0.01174
Sonnet 5.5 $0.00020 $0.00587
Haiku 4.5 $0.00010 $0.00294

Measured 12d ago against content hash eba7399c9021, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

video-reference scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.

The scan reads SKILL.md. This mod also ships 6 executable files (scripts/lib.py, scripts/reference_check.py, scripts/reference_frames.py, …), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

Origin

This is a copy

100% identical to video-reference — 0 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.

skills/video-reference/SKILL.md · 105 lines

How it starts

The opening of the file, as written. The whole thing — 105 lines — stays where its author put it; the contents beside it link to each section on GitHub.

1. 定位

本技能把一部已完成的成片拆成下一次制作可参考的方法与数值,按需在本地运行。它只写自己的测量、标注与导出物,不调用 MiMo,也不对生产流程评分或拦截。

Agent 的角色是拆片编辑:先如实记录"这部片子怎么做的"(事实),再提炼"换一部素材还能怎么用"(方法)。 两者必须分开:事实只留在本地工作目录,导出物只含方法和测得的数值。

2. 输入

  • 成片文件(.mp4 / .mov / .mkv / .webm)。
  • 对这部成片跑出的视频理解产物目录 U(建议 ASR_SEGMENT_SECONDS=5,窗口越短,旁白语速越准)。 本技能读取其中可选的 asr_result.json、asr_timing_evidence.json、background_research.json、understanding_index.json (其 characters 的名字、别名与 ASR 提及都进入人名泄漏扫描); 标注时 Agent 还应看故事板 / contact sheet 与 vlm_analysis.json。

没有理解产物也能跑,但旁白语速为空,泄漏扫描只剩 Agent 自己写的 entities,check 会给出警告; 看画面可用 frames --span 0,<时长> --step 2 生成的抽帧拼图。

下面的 scripts/... 均相对于本技能目录。

3. 流程

python3 scripts/reference.py measure <成片> --work-dir U        # 一次 ffmpeg:切点 + 响度,按文件身份缓存
python3 scripts/reference.py frames  <成片> --work-dir U --review          # 逐帧看被压下的疑似切点
python3 scripts/reference.py frames  <成片> --work-dir U --longest 5       # 看最长镜头里有没有漏切
python3 scripts/reference.py check   --work-dir U [--json]      # 校验并打印派生值
python3 scripts/reference.py export  --work-dir U --out <下次运行的 work_dir>/production_reference.json
  1. measure 写 U/reference_measurements.json(只由脚本写):镜头切点、镜长分布、每分钟切点数、10 秒切点曲线、 整体响度 / LRA / 真峰值、逐秒短期响度。它要解码整片:5 分钟 720p 约 15–70 秒,4K 约 7 分钟;成片不变时复用, 只改 --soft-score/--hard-score 不重新解码。切点规则:scdet 分数 ≥10 必算;≥4 且是前后 0.3 秒内其他帧(相邻帧除外) 两倍以上的孤立峰才算。运动镜头、急推拉、闪光会被压下,并列进 shots.review_windows;scdet 分数会减去前一帧的帧差, 从快速运动切进静止镜头的硬切可能只有 1 分,所以帧差本身的单侧峰(mafd_peaks)也进待复核窗口,但不会自动算切点。
  2. 复核切点:固定阈值在真实成片上两头都错(暗场硬切只有 5–7 分;快速运动的单个镜头每 0.2 秒一个 7–8 分的峰), 所以每次都要看图。frames --review 逐帧拼出每个待复核窗口,frames --longest 5 在最长的 5 个镜头里均匀取 12 帧; 页面在 U/reference_frames/,命令打印每格对应的秒数。看完写 labels.cut_fixes:add 漏掉的切点秒数, remove 误报的切点秒数(±0.1 秒内对上测得的切点);看过无需改动就写 {}。重新 measure 后待复核窗口数变了,旧的 cut_fixes 不再算数,按新窗口重看一遍。之后所有镜头数值、段内切点密度和导出都用复核后的切点。
  3. 标注:Agent 写 U/reference_breakdown.json 的 labels——音轨归属 audio_spans、叙事段落 sections、 字幕形态 subtitles、标注依据 basis。audio_spans 的边界放在声音实际起止处(听得到的人声起点与止点), 不放在字幕或旁白块的开始处(按字幕出现帧定的旁白结束点实测晚 0.24–0.42 秒)。U/speech_boundary_anchors.json 存在时, 用它的 acoustic_pauses(start 是人声停下处,end 是下一句开口处)对齐边界;switch_on_cut_share 只容差 ±0.25 秒, 边界放错它就只反映标注习惯。原片自带的画外音、内心独白也是原片音轨,记 original_dialogue;要区分时写进 fact 和方法。 只有 labels 时就可以跑 check,它会打印 derived(各音轨占比、段内切点密度、 旁白语速、声音切换与画面切点的对齐比例、分数位置的结构、第一次原声出现位置)。
  4. 事实与方法:在同一文件写 source_facts(带时间或测量锚点、显式 entities)和 methods (rule、applies_when、avoid_when、applies_to、evidence、targets)。targets 只写 {"from": "<测量路径>"}, 数值由 export 从当前测量填入,Agent 永远不手写数字。
  5. check → export:零 error 才导出;导出后对产物再做一次泄漏扫描。

Read the full file on GitHub · 105 lines

Files

What ships with it

7 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 12d ago Changed · -1 lines eba7399c9021
  2. 13d ago Changed · -3 lines 023fd0688f0b
  3. 24d ago First seen · 109 lines · 98 tokens per session scan A 88940900b89b

Subscribe to this mod's changes

video-reference is a skill published in the GitHub repository zenstory-ai/video-recap-skills (592 stars, last pushed 7d ago), licensed MIT. It adds 98 tokens to every session and 2,936 once invoked, about $0.0004 per session on Opus 5.5. A static security scan graded it A with 0 findings. It is 100% identical to video-reference, differing in 0 lines, and is treated as a copy.

Related

Other skills, from other repositories

narrator-ai-cli

A command-line client for creating AI-narrated movie and short-drama videos. It can guide a workflow from selecting source material, templates, music, and voices to writing narration and combining the video.

NarratorAI-Studio/narrator-ai-cli-skill · 132 tokens

narrator-ai-cli

Create AI-narrated film/drama commentary videos via CLI. Two workflow paths (Original & Adapted narration), 93 movies, 146 BGM tracks, 63 dubbing voices in 11 languages, 90+ narration templates. Use when creating narration videos, film commentary, short drama dubbing, or video production.

NarratorAI-Studio/narrator-ai-cli · 73 tokens

recut-editor

A timeline-editing skill for creating new videos or changing parts of an existing timeline in a CapCut-style editor. It can work with media assets and export the result while keeping edits undoable.

6174/recut · 78 tokens

recut-worlds

A toolset for modeling worlds and their canvases through properties, relationships, types, and visible elements. It also supports proposing media generation in the context of a world before asking for confirmation.

6174/recut · 49 tokens

generation-prompt

A reference guide for writing production-ready prompts for image and video generation models. It covers fixed visual style, typed reference anchors, multi-shot sequences, and agreements between sound and picture.

6174/recut · 82 tokens

shot

A filmmaking reference for deciding how to shoot a shot or scene, including framing, camera angle, lens choice, movement, staging, and continuity between shots.

6174/recut · 54 tokens