comfyui-research

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

ComfyUI Research Skill

ComfyUI研究技能

Keeps the VideoAgent knowledge base current by monitoring sources, extracting knowledge, and flagging stale information.

通过监控信息源、提取知识并标记过时信息，保持VideoAgent知识库的时效性。

Triggers

触发条件

On demand: User invokes
```
/research comfyui
```
or asks to check for updates
Session start: Quick staleness check (read
```
references/staleness-report.md
```
)
Before workflow generation: Targeted check if the specific model/technique is >2 months old

按需触发：用户调用
```
/research comfyui
```
指令或要求检查更新
会话启动时：快速执行过时检查（读取
```
references/staleness-report.md
```
）
工作流生成前：若特定模型/技术已超过2个月，执行针对性检查

Research Pipeline

研究流程

1. Discovery

1. 发现阶段

Search for new developments across three source types:

YouTube Channels (technique extraction):

Channel	Focus	URL
Pixaroma	ComfyUI tutorials, GGUF, beginner-friendly	youtube.com/@pixaroma
Kijai	Advanced nodes, model implementations	youtube.com/@Kijai
Sebastian Kamph	ComfyUI workflows, automation	youtube.com/@sebastiankamph
Olivio Sarikas	AI art techniques, ComfyUI	youtube.com/@OlivioSarikas
Matt Wolfe	AI tool roundups	youtube.com/@maboroshi
Nerdy Rodent	Stable Diffusion deep dives	youtube.com/@NerdyRodent

GitHub Repositories (node/model updates):

Repository	What to Check
comfyanonymous/ComfyUI	Releases, breaking changes
Kosinkadink/ComfyUI-AnimateDiff-Evolved	New motion modules
cubiq/ComfyUI_IPAdapter_plus	Maintenance status, forks
ltdrdata/ComfyUI-Impact-Pack	FaceDetailer updates
Wan-AI/Wan2.1-Preview	New model versions
hzwer/Practical-RIFE	Interpolation updates
resemble-ai/chatterbox	Voice model updates
SWivid/F5-TTS	TTS updates

HuggingFace (model releases):

Trending models:
```
huggingface.co/models?sort=trending
```
Filter: diffusers, video, audio, text-to-image
Watch for: New FLUX variants, Wan updates, identity models, voice models

在三类信息源中搜索新进展：

YouTube频道（技术提取）：

频道名称	关注方向	链接
Pixaroma	ComfyUI教程、GGUF、新手友好内容	youtube.com/@pixaroma
Kijai	高级节点、模型实现	youtube.com/@Kijai
Sebastian Kamph	ComfyUI工作流、自动化	youtube.com/@sebastiankamph
Olivio Sarikas	AI艺术技术、ComfyUI	youtube.com/@OlivioSarikas
Matt Wolfe	AI工具汇总	youtube.com/@maboroshi
Nerdy Rodent	Stable Diffusion深度解析	youtube.com/@NerdyRodent

GitHub仓库（节点/模型更新）：

仓库名称	检查内容
comfyanonymous/ComfyUI	版本发布、破坏性变更
Kosinkadink/ComfyUI-AnimateDiff-Evolved	新运动模块
cubiq/ComfyUI_IPAdapter_plus	维护状态、派生仓库
ltdrdata/ComfyUI-Impact-Pack	FaceDetailer更新
Wan-AI/Wan2.1-Preview	新模型版本
hzwer/Practical-RIFE	插值更新
resemble-ai/chatterbox	语音模型更新
SWivid/F5-TTS	TTS更新

HuggingFace（模型发布）：

热门模型：
```
huggingface.co/models?sort=trending
```
筛选条件：diffusers、video、audio、text-to-image
重点关注：新FLUX变体、Wan模型更新、身份模型、语音模型

2. Extraction

2. 提取阶段

For each discovery, extract structured knowledge:

From YouTube videos: Use

youtube-video-analyst

youtube-chapter-clipper

skills to:

Get transcript
Identify new models mentioned (name, source, requirements)
Identify new techniques (workflow changes, parameter discoveries)
Identify new nodes (package name, what it does)
Note any benchmark results or comparisons

From GitHub releases:

Check release notes for breaking changes
Identify new features or supported models
Note version requirements
Check if deprecated features affect our workflows

From HuggingFace:

Model card: architecture, training data, requirements
VRAM requirements and quantization options
ComfyUI compatibility (is there a node?)
Quality comparison vs current recommendations

针对每个新发现，提取结构化知识：

从YouTube视频中提取： 使用

youtube-video-analyst

或

youtube-chapter-clipper

技能完成以下操作：

获取视频字幕
识别提及的新模型（名称、来源、运行要求）
识别新技术（工作流变更、参数新发现）
识别新节点（包名称、功能）
记录基准测试结果或对比信息

从GitHub版本发布中提取：

查看发布说明中的破坏性变更
识别新功能或支持的模型
记录版本要求
检查已弃用功能是否会影响现有工作流

从HuggingFace中提取：

模型卡片：架构、训练数据、运行要求
VRAM要求与量化选项
ComfyUI兼容性（是否有对应节点）
与当前推荐模型的质量对比

3. Integration

3. 整合阶段

Update reference files with timestamped entries:

markdown

<!-- Updated: 2026-02-06 | Source: Pixaroma Ep 45 -->

为参考文件添加带时间戳的条目：

markdown

<!-- Updated: 2026-02-06 | Source: Pixaroma Ep 45 -->

New Model: XYZ

新模型：XYZ

Download: {url}
VRAM: {requirement}
Notes: {key findings}


**Which file to update:**
| Finding Type | Update File |
|-------------|-------------|
| New model | `references/models.md` |
| New workflow technique | `references/workflows.md` |
| LoRA training discovery | `references/lora-training.md` |
| Voice tool update | `references/voice-synthesis.md` |
| New prompt technique | `references/prompt-templates.md` |
| Bug workaround | `references/troubleshooting.md` |
| Model landscape shift | `foundation/model-landscape.md` |

下载地址：{url}
VRAM要求：{requirement}
说明：{核心发现}


**对应更新的文件：**
| 发现类型 | 更新文件 |
|-------------|-------------|
| 新模型 | `references/models.md` |
| 新工作流技术 | `references/workflows.md` |
| LoRA训练新发现 | `references/lora-training.md` |
| 语音工具更新 | `references/voice-synthesis.md` |
| 新提示词技术 | `references/prompt-templates.md` |
| 问题解决方法 | `references/troubleshooting.md` |
| 模型格局变化 | `foundation/model-landscape.md` |

4. Staleness Audit

4. 过时审计

Scan all reference files for

<!-- Updated: -->

timestamps:

Thresholds:

Category	Max Age	Action
Models	3 months	Flag: "Model recommendations may be outdated"
Nodes	2 months	Flag: "Custom node versions may have updates"
Techniques	6 months	Flag: "Techniques may have been superseded"
Voice tools	3 months	Flag: "Voice synthesis landscape may have changed"

Generate report: Update

references/staleness-report.md

with current status.

扫描所有参考文件中的

<!-- Updated: -->

时间戳：

阈值设置：

类别	最长有效期	操作
模型	3个月	标记：“模型推荐可能已过时”
节点	2个月	标记：“自定义节点版本可能存在更新”
技术	6个月	标记：“技术可能已被替代”
语音工具	3个月	标记：“语音合成领域可能已发生变化”

生成报告：更新

references/staleness-report.md

以反映当前状态。

Research State

研究状态

Maintained in

skills/comfyui-research/state/

state/
  last-research-run.json    # When, what was checked, what was found
  watch-list.yaml           # Sources to monitor (see references/)

状态信息存储在

skills/comfyui-research/state/

目录下：

state/
  last-research-run.json    # 上次运行时间、检查内容、发现成果
  watch-list.yaml           # 监控的信息源（详见参考文件）

last-research-run.json

json

{
  "last_run": "2026-02-06T12:00:00Z",
  "sources_checked": 15,
  "findings": 3,
  "updates_made": [
    {"file": "models.md", "entry": "Added ModelX", "source": "HuggingFace"},
    {"file": "workflows.md", "entry": "Updated Wan settings", "source": "Pixaroma Ep 50"}
  ],
  "stale_entries": 0
}

json

{
  "last_run": "2026-02-06T12:00:00Z",
  "sources_checked": 15,
  "findings": 3,
  "updates_made": [
    {"file": "models.md", "entry": "Added ModelX", "source": "HuggingFace"},
    {"file": "workflows.md", "entry": "Updated Wan settings", "source": "Pixaroma Ep 50"}
  ],
  "stale_entries": 0
}

Quick Staleness Check (Session Start)

快速过时检查（会话启动时）

Fast check without full research:

Read
```
references/staleness-report.md
```
If
```
last_run
```
is >2 weeks ago:
- Notify: "Research data is {N} days old. Run
```
/research comfyui
```
  for updates."
If any entries are beyond their threshold:
- Notify: "Stale items found: {list}. These may have newer alternatives."

无需完整研究的快速检查：

读取
```
references/staleness-report.md
```
若
```
last_run
```
已超过2周：
- 通知用户：“研究数据已过期{N}天。请运行/research comfyui获取更新。”
若存在超过阈值的条目：
- 通知用户：“发现过时内容：{列表}。这些内容可能已有更新的替代方案。”

Full Research Run

完整研究运行

When user triggers

/research comfyui

Check all YouTube channels for recent uploads (last 30 days)
Check all GitHub repos for new releases
Check HuggingFace trending for relevant models
Extract findings using youtube-video-analyst
Update reference files with timestamped entries
Run staleness audit on all references
Update staleness report
Summarize findings for user

当用户触发

/research comfyui

时：

检查所有YouTube频道的近期上传内容（过去30天）
检查所有GitHub仓库的新版本发布
检查HuggingFace热门模型中的相关内容
使用youtube-video-analyst提取研究成果
为参考文件添加带时间戳的条目
对所有参考文件执行过时审计
更新过时报告
为用户总结研究成果

Targeted Research

针对性研究

When asked about a specific topic:

Search YouTube for recent videos on the topic
Search HuggingFace for related models
Check GitHub for related node packages
Synthesize findings
Update relevant reference files if significant discoveries

当用户询问特定主题时：

在YouTube搜索该主题的近期视频
在HuggingFace搜索相关模型
在GitHub搜索相关节点包
整合研究成果
若有重大发现，更新对应参考文件

Integration with Other Skills

与其他技能的整合

Skill	How Research Helps
`comfyui-workflow-builder`	Ensures recommended models/nodes are current
`comfyui-character-gen`	Updates identity method recommendations
`comfyui-video-pipeline`	Flags new video models (Wan updates, new engines)
`comfyui-voice-pipeline`	Tracks voice synthesis tool releases
`comfyui-troubleshooter`	Adds community-discovered workarounds

技能	研究成果的作用
`comfyui-workflow-builder`	确保推荐的模型/节点为当前最新版本
`comfyui-character-gen`	更新身份生成方法的推荐
`comfyui-video-pipeline`	标记新的视频模型（Wan模型更新、新引擎）
`comfyui-voice-pipeline`	追踪语音合成工具的发布
`comfyui-troubleshooter`	补充社区发现的问题解决方法

Reference

参考文件

```
references/staleness-report.md
```
- Current freshness status
```
references/evolution.md
```
- Update protocol and changelog

skills/comfyui-research/references/watch-list.yaml

- Monitored sources

```
references/staleness-report.md
```
- 当前内容新鲜度状态
```
references/evolution.md
```
- 更新协议与变更日志

skills/comfyui-research/references/watch-list.yaml

- 监控的信息源