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 IvanYangYangXi/artclaw_bridge --skill comfyui-img2imggit clone --depth 1 https://github.com/IvanYangYangXi/artclaw_bridgeWrote 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/ivanyangyangxi/artclaw_bridge/comfyui-img2img)<a href="https://agentmods.dev/skills/ivanyangyangxi/artclaw_bridge/comfyui-img2img"><img src="https://agentmods.dev/badge/skills/ivanyangyangxi/artclaw_bridge/comfyui-img2img/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/ivanyangyangxi/artclaw_bridge/comfyui-img2img"><img src="https://agentmods.dev/badge/skills/ivanyangyangxi/artclaw_bridge/comfyui-img2img.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 | $0.00064 | $0.02950 |
| Opus 5 | $0.00032 | $0.01475 |
| Sonnet 5 | $0.00013 | $0.00590 |
| Haiku 4.5 | $0.00006 | $0.00295 |
Grade A, and why
comfyui-img2img 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.
How it starts
The opening of the file, as written. The whole thing — 390 lines — stays where its author put it; the contents beside it link to each section on GitHub.
ComfyUI 图生图 (img2img)
基于参考图像生成新图像,支持风格迁移、图像增强、局部重绘等。
预注入变量
run_python 已注入以下变量,直接使用,无需 import:
| 变量 | 类型 | 说明 |
|---|---|---|
S |
[] |
无选中概念 |
W |
None |
无当前文件概念 |
L |
ComfyUI Lib | L.nodes, L.folder_paths, L.model_management |
nodes |
module | 节点注册表 |
folder_paths |
module | 模型/输出路径管理 |
client |
ComfyUIClient | HTTP API 客户端 |
submit_workflow |
func | 提交 workflow 并等待完成 |
save_preview |
func | 保存图片并输出 [IMAGE:] 标记 |
核心工作流
标准的 img2img 流程:
LoadImage → CLIPTextEncode(×2) → CheckpointLoader → KSampler
→ VAEDecode → SaveImage
↑
VAEEncode (将输入图编码为 latent)
使用方法
方法 1: 使用便捷函数 (推荐)
# 查询可用模型
ckpts = folder_paths.get_filename_list("checkpoints")
print(f"可用模型: {ckpts[:5]}")
# 构建 img2img workflow
wf = {}
# 1. 加载输入图像
wf["1"] = {
"class_type": "LoadImage",
"inputs": {"image": "input_image.png"} # 图片需在 input 目录
}
# 2. 加载模型
wf["2"] = {
"class_type": "CheckpointLoaderSimple",
"inputs": {"ckpt_name": "sd_xl_base_1.0.safetensors"}
}
# 3. 编码提示词
wf["3"] = {
"class_type": "CLIPTextEncode",
"inputs": {"text": "masterpiece, best quality, a beautiful landscape", "clip": ["2", 1]}
}
wf["4"] = {
"class_type": "CLIPTextEncode",
"inputs": {"text": "low quality, blurry, ugly", "clip": ["2", 1]}
}
# 4. 将输入图编码为 latent
wf["5"] = {
"class_type": "VAEEncode",
"inputs": {"pixels": ["1", 0], "vae": ["2", 2]}
}
# 5. KSampler (关键:denoise < 1.0)
wf["6"] = {
"class_type": "KSampler",
"inputs": {
"model": ["2", 0],
"positive": ["3", 0],
"negative": ["4", 0],
"latent_image": ["5", 0],
"seed": 42,
"steps": 20,
"cfg": 7.0,
"sampler_name": "euler",
"scheduler": "normal",
"denoise": 0.75 # 关键参数:0.0=原图, 1.0=完全重绘
}
}
# 6. 解码
wf["7"] = {
"class_type": "VAEDecode",
"inputs": {"samples": ["6", 0], "vae": ["2", 2]}
}
# 7. 保存
wf["8"] = {
"class_type": "SaveImage",
"inputs": {"images": ["7", 0], "filename_prefix": "img2img_result"}
}
# 提交执行
result = submit_workflow(wf)
print(f"生成完成: {result}")
# 展示结果
if result.get("images"):
save_preview(result["images"][0])
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 First seen · 390 lines · 64 tokens per session scan A 8f7a406d8f41
comfyui-img2img is a skill published in the GitHub repository IvanYangYangXi/artclaw_bridge (35 stars, last pushed 4mo ago), licensed MIT. It adds 64 tokens to every session and 2,950 once invoked, about $0.0003 per session on Opus 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 2026-08-30.
Other skills, from other repositories
dcc-mcp-core
Foundation library for the DCC Model Context Protocol (MCP) ecosystem. Provides Rust-powered action management, skills system, IPC transport, MCP Streamable HTTP server (2025-03-26 spec, with 2025-06-18 and 2025-11-25 awareness), sandbox security, shared memory, screen capture, USD scene support, and telemetry for…
dcc-cua
A routing guide for controlling application interfaces through the project’s DCC-CUA system. DCC applications are digital-content tools such as Maya, Blender, and Houdini.
marketplace-publish-extension
Infrastructure skill — publish (register/update) an extension package to a marketplace catalog. Reads the extension's SKILL.md frontmatter, constructs a CatalogEntry, and upserts it into the target marketplace.json. Optionally commits and pushes when the catalog source is a git repository. Use after scaffolding an…
clawhub-compat
Example skill — demonstrates full compatibility with the ClawHub/OpenClaw skill format. Use as a reference when creating skills for both the dcc-mcp-core ecosystem and ClawHub marketplace. Not intended for production use — this is an authoring reference only.
example-layered-skill
Example skill — reference implementation of the internal layered architecture pattern (Tools / Services / Utils) for complex skills with shared business logic. Use as a template when a skill outgrows a single scripts/execute.py file. Not intended for production use — see docs/guide/skills.md for the architectural…
marketplace-create-extension
Infrastructure skill — scaffold a new marketplace extension package (SKILL.md + tools.yaml + scripts/) with MIT-0 licensing. Use when creating a publishable marketplace entry for any DCC host. Not for editing existing extensions or driving live DCC scenes — use domain skills for that.