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 sd-operation-rulesgit 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/sd-operation-rules)<a href="https://agentmods.dev/skills/ivanyangyangxi/artclaw_bridge/sd-operation-rules"><img src="https://agentmods.dev/badge/skills/ivanyangyangxi/artclaw_bridge/sd-operation-rules/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/sd-operation-rules"><img src="https://agentmods.dev/badge/skills/ivanyangyangxi/artclaw_bridge/sd-operation-rules.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 | $0.00061 | $0.02577 |
| Opus 5 | $0.00030 | $0.01288 |
| Sonnet 5 | $0.00012 | $0.00515 |
| Haiku 4.5 | $0.00006 | $0.00258 |
Grade A, and why
sd-operation-rules scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 11d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 250 lines — stays where its author put it; the contents beside it link to each section on GitHub.
SD 操作规则
📋 通用节点图工作流（伪代码→映射→分组创建→验证→布局）见
dcc-node-graph-workflowskill。 本 skill 专注 SD 的 API 约束、截图验证流程和致命陷阱。
🔴 代码拆分规则
每次 run_python 调用做一个逻辑步骤（通常 3-5 个节点操作）。
避免：
- ⛔ 一次调用创建整个图（10+ 节点 + 全部设参 + 全部连线）→ 中间报错后续全丢
- ⛔ 截图和预写分析放同一次调用 → 分析不是基于真实观察
推荐模式：
调用1: 创建 mesh_1 + levels，设参连线
调用2: save_preview(levels) ← 只截图，不写分析
→ 看到图后分析 → 通过则继续
调用3: 创建 transformation × 2 + blend
调用4: save_preview(blend) → 分析 → 通过则继续
🔴 参数设置规则
⛔ 禁止猜测参数名和类型！创建节点后第一步永远是动态查询。
# ✅ 正确：先查询，再设值
node = graph.newNode("sbs::compositing::levels")
for p in node.getProperties(SDPropertyCategory.Input):
pid = p.getId()
if pid.startswith("$"): continue
v = node.getPropertyValue(p)
if v is None: print(f" PORT: {pid}") # 连接端口
else: print(f" PARAM: {pid} = {v.get()} ({type(v).__name__})")
已知陷阱：
| 节点 | 参数名 | 正确类型 | 常见错误 |
|---|---|---|---|
| levels | levelinlow/levelinhigh/levelinmid |
Float4 | 猜成 inlow + Float |
| blend | opacitymult (标量) |
SDValueFloat | 猜成 opacity（那是连接端口） |
| uniform | colorswitch |
SDValueBool | ⛔ 不设 = 输出永远灰色 |
| 库节点 Enum | — | SDValueInt.sNew(n) | 无 SDValueEnum.sNew |
🔴 截图驱动的迭代循环
截图两步法（⛔ 必须分两次调用）
步骤 1 (run_python): 只截图
save_preview(node, "CP1_structure")
步骤 2 (看到图后): 逐维度分析，写出判断
4 个必须分析的维度
| 维度 | 检查点 | 问题信号 |
|---|---|---|
| 亮度/对比度 | 是否有足够动态范围？ | 全黑→立即向上游追查 |
| 图案正确性 | 是否符合预期？ | artifact/意外图案 |
| 密度/比例尺 | 基于物理尺度声明检查 | 太稀/太密 → 调 scale |
| CG 感 | 看起来是否过于规则？ | 需要 warp/噪波打破 |
分析输出格式：
[CP1] 分析 (attempt N/3):
亮度: 正常/全黑/...
图案: 正确/有artifact/...
密度: 合理/过密/...
CG感: 无/有...
判定: ✅通过 / ⛔需修复
修复: [根因 → 具体改法]
修复循环（⛔ 最多 3 次，超出则报告并停止）：
截图 → 分析 → 有问题?
→ 向上游逐节点截图追查 → 找第一个出问题的节点
→ 修复 → 重新截图验证 → retry_count++
→ retry_count >= 3? 停！汇报，请求人工指导
🔴 参考工程优先原则
⛔ 禁止从零构建功能模块！先查参考工程有没有可复用的。
| 优先级 | 方式 | 示例 |
|---|---|---|
| 1⃣ 加载参考工程子图 | loadUserPackage → newInstanceNode | Creases_Filter、fabric_weathering |
| 2⃣ 复制参考工程节点链参数 | 按参考工程参数值重建 | fabric_009 的 mesh_1+transformation 编织链 |
| 3⃣ 复用参考工程管线架构 | 替换纹理源，保留管线结构 | fabric_009 的 5层着色架构 |
| 4⃣ 自己构建 | 仅当参考工程完全没有时 | — |
What ships with it
2 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.
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.
- 11d ago First seen · 250 lines · 61 tokens per session scan A ad42ae715674
sd-operation-rules is a skill published in the GitHub repository IvanYangYangXi/artclaw_bridge (35 stars, last pushed 4mo ago), licensed MIT. It adds 61 tokens to every session and 2,577 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…
maya-pipeline
Domain skill — Maya asset pipeline orchestration: set up project directory structures, export scenes to USD, and coordinate multi-step DCC workflows. Use when initialising a Maya project or exporting assets for a downstream pipeline. Not for raw geometry editing — use maya-geometry for that. Not for low-level USD file…
imagemagick-tools
Infrastructure skill — image processing and manipulation via ImageMagick: resize, composite, convert formats, add watermarks. Use when batch-processing textures, thumbnails, or rendered images at the file level. Not for in-DCC texture or material editing — use a domain skill bound to the specific DCC for that.
asset-source
Gateway skill for cross-DCC asset import — search and resolve assets into a validated AssetDescriptor (local path + attribution). Demo source returns static catalog entries; production sources can add download or remote resolution without changing the contract.
ffmpeg-media
Infrastructure skill — media conversion and processing via FFmpeg: convert video/audio formats, extract frames, resize, and transcode. Use when manipulating raw media files (mp4, mov, wav, image sequences) regardless of DCC context. Not for DCC-specific render output handling — use a domain pipeline skill for…
maya-geometry
Domain skill — Maya geometry primitives: create spheres, cubes, cylinders; bevel, extrude, and merge polygon components. Use for individual geometry creation or editing operations inside Maya. Not for full asset export pipelines — use maya-pipeline for that. Not for USD scene inspection — use usd-tools for that.