Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/DDDFXYqiming/Agent_Extensionsnpx agentmods add skills/dddfxyqiming/agent_extensions/video-notes-generatorWrote 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/dddfxyqiming/agent_extensions/video-notes-generator)<a href="https://agentmods.dev/skills/dddfxyqiming/agent_extensions/video-notes-generator"><img src="https://agentmods.dev/badge/skills/dddfxyqiming/agent_extensions/video-notes-generator/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/dddfxyqiming/agent_extensions/video-notes-generator"><img src="https://agentmods.dev/badge/skills/dddfxyqiming/agent_extensions/video-notes-generator.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00051 | $0.02195 |
| Opus 5.5 | $0.00020 | $0.00878 |
| Sonnet 5.5 | $0.00010 | $0.00439 |
| Haiku 4.5 | $0.00005 | $0.00219 |
Grade A, and why
video-notes-generator 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 19d 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.
How it starts
The opening of the file, as written. The whole thing — 89 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Video notes generator
脚本负责获取、完整转写分块和本地视觉候选筛选；当前 Agent 负责原生读图、按需补查和综合笔记。脚本不调用总结模型、不自动安装依赖。
来源与更新
| 项 | 值 |
|---|---|
| 源仓库 | https://github.com/DDDFXYqiming/Agent_Extensions |
| 仓库内路径 | General_skills/video-notes-generator |
| 本地 clone | C:\Users\39795\Desktop\AI_project\Agent_Extensions |
| 运行副本 | ~/.dsh/skills/video-notes-generator；若本机装了 codex 客户端，同步到 ~/.codex/skills/video-notes-generator |
| 同步方式 | 改源树后 robocopy /E 到运行副本（先备份运行副本 .env），不要直接改运行副本 |
更新流程：源仓库 git pull（main 跟踪 origin/main）→ 改动落在 General_skills/video-notes-generator → robocopy 同步 → 同步 codex 侧 → 校验两边文件 MD5 一致。运行副本里若有 .env 等本地配置，同步前备份、同步后还原。
获取素材
python <skill>/scripts/video_to_notes.py "<video-url-or-local-file>" -o ./notes
字幕优先，无字幕使用 faster-whisper（默认 default 档）或 whisper.cpp。正常任务保留视觉采样；下载/抽帧失败可重试一次，仍失败明确报告证据缺口。--no-frames 仅用于用户指定纯文本或明确报告的降级。
转写模型用档位而非具体模型名：WHISPER_MODEL=default（或省略）为日常档，WHISPER_MODEL=high 用于重要或嘈杂语音。档位映射集中在 runtime config/whisper_models.json 的 tiers 键，脚本按序尝试并在失败时沿候选链自动降级、在 stderr 与 source_note 报告实际模型。选用 high 档前检查映射文件的 tiers_updated：超过 6 个月或文件缺失时，先联网检索 faster-whisper 当前推荐模型、更新 tiers 与日期再运行；检索结果只写映射文件，不改本文件。发生降级时终稿必须披露实际使用的模型。
每个视频使用独立输出目录。已有 manifest 时补看不必重跑下载转写。重跑素材步骤不覆盖 Agent 终稿；源文件变化后须重新核对旧终稿。
理解与补查
- 读
*_transcript_chunks.md和*_visual_manifest.json。前者是完整带时间戳原文，不是摘要；短视频读全部，长视频逐块覆盖全部并保存真正的片段摘要，不得只读每块开头。*_draft_notes.md仅为导航，终稿由 Agent 创建。 - 实际查看候选帧或带帧号/时间戳的
contact_sheets，建立全片视觉概览。直接使用 manifest 返回的完整路径，避免手工重写长路径。拼图用于概览；文字、公式、图表数据和 UI 细节需打开单帧。没有 Pillow 就直接看候选帧。空转写时先看视觉概览，必要时只做一次针对性的音频复核，不反复转写无语音的演示。 - 按主题或操作片段联合阅读少量相关图片和对应转写。允许多帧加附近文本，不设“一次只能一帧”的限制。记录讲解、独立视觉事实、关系和疑点；画面独有信息可进入正文，不必强求字幕也说过。单张截图不能证明未观察到的动作。
- 关键小字、前后状态跳变、字幕画面冲突、无声变化片段需要按需补看。快速切换的案例/模板蒙太奇即使没有大时间空档，也应按事件密度局部加密。原下载清晰度不足时报告限制或获取高清来源；放大低清图不能恢复细节。OCR 可辅助密集文字，但不能代替原图核对。
- 关键疑点解决或补看没有新增重要信息时停止。初始候选默认上限24张；补看可先在12张内安排，按信息密度追加。预算用尽仍有关键缺口应披露，不能以模型信心替代证据。
局部补看，无需重跑下载转写
python <skill>/scripts/video_to_notes.py --review-manifest ./notes/ID_visual_manifest.json --timestamps 12,13,14
# 源像素 x,y,width,height；另保留全帧提供位置上下文
python <skill>/scripts/video_to_notes.py --review-manifest ./notes/ID_visual_manifest.json --timestamps 13 --crop 100,80,800,500
What ships with it
17 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.
- CHANGELOG.md 4.4 KB
- references/bilibili-412-api-fallback.md 2.1 KB
- references/bilibili-uploader-discovery.md 4.0 KB
- references/bilibili-uploader-exact-space-api.md 3.5 KB
- references/bilibili-windows-ssl-and-long-audio.md 2.8 KB
- references/bulk-visual-enrichment.md 849 B
- references/multimodal-evidence.md 2.0 KB
- references/platform-support.md 1.9 KB
- references/prompt-templates.md 1.1 KB
- references/style-guide.md 3.7 KB
- scripts/install_deps.sh 2.6 KB runs code
- scripts/test_video_materials.py 5.8 KB runs code
- scripts/video_to_notes.py 67 KB runs code
- scripts/visual_sampling.py 10.0 KB runs code
- templates/.env.example 1.4 KB
- templates/config.yaml 732 B
- templates/whisper_models.json 695 B
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.
- 19d ago Changed · +17 lines 03dc707cab27
- 1mo ago Changed · -55 lines · -15 tokens per session 823847ce933e
- 1mo ago Changed · +6 lines 971e3937b6af
- 2mo ago First seen · 121 lines · 66 tokens per session scan A e4eeaaae1411
video-notes-generator is a skill published in the GitHub repository DDDFXYqiming/Agent_Extensions (6 stars, last pushed 20d ago), licensed MIT. It adds 51 tokens to every session and 2,195 once invoked, about $0.0002 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 2026-08-31.
Other skills, from other repositories
wayfinder
A planning method that maps a large, uncertain project into linked decision tickets in a repository's issue tracker. An issue tracker is a system for recording and organizing project work.
setup-matt-pocock-skills
A one-time setup guide for configuring a repository so other engineering skills know its issue tracker, triage labels, domain documentation, and architecture-decision locations.
setup-ts-deep-modules
A TypeScript repository setup that keeps each package's internal code private behind root-level entry files. A package is a reusable unit of code, and an entry point is the file other code is meant to import.
codebase-design
A shared vocabulary and design guide for shaping code modules: small interfaces hide substantial behavior, with clear seams where behavior can be changed or tested.
triage
A process for reviewing issues and outside pull requests through category and status labels. An issue is a recorded problem or request, while a pull request is a proposed code change for review.
pr
A writing guide for creating pull request descriptions, which explain a proposed code change to reviewers. It uses sections for the summary, evidence, merge risk, and the likely area affected by the change.