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.
npx skills add zenstory-ai/oh-story-dsh --skill video-understandinggit clone --depth 1 https://github.com/zenstory-ai/oh-story-dshWrote 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.
[](https://agentmods.dev/skills/zenstory-ai/oh-story-dsh/video-understanding)<a href="https://agentmods.dev/skills/zenstory-ai/oh-story-dsh/video-understanding"><img src="https://agentmods.dev/badge/skills/zenstory-ai/oh-story-dsh/video-understanding/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.
<a href="https://agentmods.dev/skills/zenstory-ai/oh-story-dsh/video-understanding"><img src="https://agentmods.dev/badge/skills/zenstory-ai/oh-story-dsh/video-understanding.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00126 | $0.02455 |
| Opus 5.5 | $0.00050 | $0.00982 |
| Sonnet 5.5 | $0.00025 | $0.00491 |
| Haiku 4.5 | $0.00013 | $0.00246 |
Grade A, and why
video-understanding 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.
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.
Copies of this mod
1 near-identical copy found in the catalogue:
- video-understanding — 100% identical, 0 lines differ
How it starts
The opening of the file, as written. The whole thing — 107 lines — stays where its author put it; the contents beside it link to each section on GitHub.
1. 定位
本技能把源视频转成 Agent 与下游阶段可读取的理解索引，以素材观察员 / 场记的角色记录证据：
- 先观察，再解释；事实与推断分开。
- 除了“发生了什么”，还要让下游看见知识、权力、目标、关系或情绪在哪一刻变化。
- 标出由谁的 POV 承载变化、哪个反应或表演不可替代，以及哪里存在完整台词/动作的自然剪辑边界。
- 证据不足时保留不确定性，不制造戏剧结论。
2. 处理阶段
- 场景检测：写
scenes.json，包含切点、时长和废片段过滤结果。 - 抽帧：为视觉分析提取代表帧。
- ASR：通过
mimo-v2.5-asr写粗分段对白asr_result.json，并写asr_timing_evidence.json说明可用性、有限时间精度与文本修正来源。 - 静音检测：写
silence_periods.json，标注安静窗口与has_speech。 - VLM 观察：写
vlm_analysis.json，包含场景描述、深层分析和frame_facts。 - 时间线融合与创作 brief：写
timeline_fusion.json、asr_writing_chunks.json和agent_narration_brief.md。
各阶段只有在输出产物与 provenance sidecar 同时匹配当前视频及影响结果的设置时才会复用；--force 强制重算。
3. 环境要求
# ffmpeg: brew install ffmpeg | apt install ffmpeg | choco install ffmpeg
export MIMO_API_KEY=***
ASR 使用 mimo-v2.5-asr；VLM 使用 mimo-v2.5。--skip-asr 可跳过对白转写，但完整理解仍需要 MIMO_API_KEY 运行 VLM。--mimo-video-overview 可开启按场景块的视频概览。未设置 key 时重跑会复用已缓存的转写、画面分析、概览与故事索引（key 决定的默认 endpoint 不参与比对）；需要请求模型的 consolidation 记为 skipped_no_key，不发请求。缓存对不上（例如复制 work_dir 时没保留文件时间）而已有转写时，运行停下并保留转写：用 cp -p / cp -Rp / rsync -t 保留时间重新复制，或设置 key 后重跑（会重新转写）。不要用 --skip-asr 绕过，它会把现有转写替换成 []。
若 work_dir/background_research.json 存在，本技能会把剧情梗概和角色名折入 VLM 上下文；--context 可补充一条简短提示。
远程调用与本地读取
ASR 把分窗音频发给 MiMo ASR；逐场景 VLM 把代表帧、帧时间和分析提示发给 MiMo；启用 --mimo-video-overview 时，还会把按场景截取的视频片段发给 MiMo 视频模型。故事索引汇总和可选 ASR 清洗发送的是画面描述、转写与背景资料文字。
请求使用 MiMo 的 chat 接口。MIMO_API_URL 可指定地址；ASR 与视频概览可分别用 MIMO_ASR_API_URL、MIMO_VIDEO_API_URL 覆盖，凭据同样支持各自的 key。未覆盖时，按量 key 使用 https://api.xiaomimimo.com/v1，Token Plan key 使用 MIMO_TOKEN_PLAN_CLUSTER 选择的集群。逐场景帧分析与汇总使用 MIMO_API_KEY。
--skip-asr 只关闭音频转写，--no-consolidate 只关闭索引汇总；它们不会关闭 VLM。需要只在本地处理已有产物时，使用 --brief-only 或 --edited-storyboard-only，这两条路径不调用 API。
下面的 scripts/... 均相对于本技能目录。若执行器从仓库根目录启动，请给脚本路径加上本技能的绝对目录。
4. 运行命令
python3 scripts/understand.py <video> --work-dir <work_dir> [选项]
| 选项 | 默认 | 作用 |
|---|---|---|
<video> |
必填 | 源视频 |
--work-dir |
必填 | 产物目录；不存在时创建 |
--context "..." |
空 | 补充给 VLM 的简短上下文（节目名、角色名），与 background_research.json 合并 |
--scene-threshold |
0.1 |
场景检测阈值 |
--style |
纪录片 |
写进创作简报的解说风格 |
--edit-mode full|cut |
不设 | 写进简报的 recap 模式；cut 时按剪后时长估算旁白预算，已有 edited_source.mp4 时句末锚点改用剪后时间 |
--target-duration |
不设 | 写进简报的 cut 目标时长；尚无 clip_plan_validated.json 时用它估算旁白预算 |
--skip-asr |
关 | 不转写对白，把 asr_result.json 写成 []（已有转写会被覆盖），ASR 证据标为显式跳过 |
--mimo-video-overview |
关 | 按场景块运行 MiMo 视频概览，并作为逐场景主描述 |
--force |
关 | 忽略缓存，全部重算 |
--brief-only |
关 | 只用现有产物重建 agent_narration_brief.md，不抽帧、不调 API |
--edited-storyboard-only |
关 | 只按 clip_plan_validated.json 写剪后时间线 storyboard/edited_storyboard.*，并在已有的 agent_narration_brief.md 顶部加 storyboard 指引；多源计划（带 sources）从各来源 source_work_dir 的 frames/ 按其 frames_manifest.json 的 fps 取帧，tile 标 S1/S2…；不抽帧、不调 API，故事板生成失败只记日志。与 --brief-only 互斥 |
--consolidate / --no-consolidate |
开 | 生成全局故事索引 understanding_index.* |
--consolidate-asr |
关 | 另外清洗 ASR 文本，写 asr_clean.json |
What ships with it
24 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.
- references/data-schema.md 6.1 KB
- references/prompt-templates.md 1.7 KB
- references/research-guide.md 2.8 KB
- scripts/agent_text.py 7.2 KB runs code
- scripts/asr_timing_evidence.py 9.7 KB runs code
- scripts/asr.py 11 KB runs code
- scripts/briefing/__init__.py 101 B runs code
- scripts/briefing/builder.py 28 KB runs code
- scripts/briefing/context.py 14 KB runs code
- scripts/briefing/inputs.py 8.9 KB runs code
- scripts/briefing/timeline.py 18 KB runs code
- scripts/consolidate.py 27 KB runs code
- scripts/detect.py 25 KB runs code
- scripts/extract.py 2.5 KB runs code
- scripts/index_normalize.py 14 KB runs code
- scripts/lib.py 23 KB runs code
- scripts/storyboard.py 24 KB runs code
- scripts/timeline_fusion.py 3.0 KB runs code
- scripts/understand.py 192 B runs code
- scripts/understanding_brief.py 8.6 KB runs code
- scripts/understanding_cache.py 12 KB runs code
- scripts/understanding_runner.py 14 KB runs code
- scripts/understanding_storyboard.py 13 KB runs code
- scripts/vlm.py 28 KB runs code
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.
- 12d ago Changed · +4 lines e0a9612667bf
- 24d ago Changed · +25 lines e62600a4194f
- 29d ago Changed · +2 lines 3cd5bc643ea2
- 1mo ago First seen · 76 lines · 126 tokens per session scan A 86c395f032d9
video-understanding is a skill published in the GitHub repository zenstory-ai/oh-story-dsh (629 stars, last pushed today), licensed MIT. It adds 126 tokens to every session and 2,455 once invoked, about $0.0005 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-05.
Other skills, from other repositories
story-deslop
A writing editor for Chinese web novels that detects writing patterns often associated with AI-generated text and makes the prose feel more natural.
story-long-analyze
A long-form fiction analysis workflow for breaking down a novel’s opening chapters, characters, pacing, turning points, relationships, and overall structure.
browser-cdp
Use this skill when you need to control a Chrome browser via CDP (Chrome DevTools Protocol) to reuse existing login sessions. Covers: launching Chrome in debug mode, opening URLs, waiting for page load, evaluating JavaScript, taking snapshots, and extracting auth tokens. Trigger phrases: browser automation, CDP…
workflow-manager
A writing workflow tracker for chapter-based projects. It records whether each chapter is being assembled, drafted, reviewed, confirmed, polished, or archived.
review-chapter
A six-part review of a manuscript chapter. It compares the chapter with the outline and reports quality, coverage, review status, and problems supported by evidence from the text.
revise-span
A revision workflow for changing selected passages in a manuscript. It shows the original and proposed text, then applies changes only after the author confirms them.