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 agentmods add skills/dddfxyqiming/agent_extensions/vision-skillnpx skills add DDDFXYqiming/Agent_Extensions --skill vision-skillgit clone --depth 1 https://github.com/DDDFXYqiming/Agent_ExtensionsWrote 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/vision-skill)<a href="https://agentmods.dev/skills/dddfxyqiming/agent_extensions/vision-skill"><img src="https://agentmods.dev/badge/skills/dddfxyqiming/agent_extensions/vision-skill.svg" alt="Measured on agentmods" 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 | $0.00047 | $0.01140 |
| Opus 5 | $0.00023 | $0.00570 |
| Sonnet 5 | $0.00009 | $0.00228 |
| Haiku 4.5 | $0.00005 | $0.00114 |
Grade A, and why
vision-skill 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 today.
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 — 72 lines — stays where its author put it; the contents beside it link to each section on GitHub.
识图技能
当主模型不支持直接读取图片时，图片不会进入对话上下文，但用户消息中通常会附带图片的本地路径（形如剪贴板截图或附件路径）。
配置（首次使用）
-
复制
templates/.env.example为技能目录下的.env（脚本启动时自动加载，无需额外依赖），填入：VISION_API_URL：视觉模型 OpenAI 兼容完整接口地址（含路径，如https://api.example.com/v1/chat/completions）VISION_MODEL：模型名VISION_API_KEY：API Key
也可以直接导出同名环境变量；环境变量优先级高于
.env文件（便于 CI/容器注入）。 -
运行
python scripts/vision.py --check自检，确认配置生效。
使用步骤
-
从用户消息中找到图片路径；路径不明确时先找最近的剪贴板截图：
Get-ChildItem $env:TEMP\codex-clipboard-*.png | Sort-Object LastWriteTime -Descending | Select-Object -First 3 FullName,Length -
运行脚本识别图片：
cd <技能目录>; python scripts/vision.py "<图片绝对路径>" "（可选）具体识图要求" -
基于脚本输出的文字描述回答用户；描述中的重要文字、报错信息要原样转述。
常用选项（按场景选择）
| 场景 | 命令 |
|---|---|
| 一般识图 | vision.py "<图>" |
| 提取所有文字（OCR，保持排版） | vision.py "<图>" --mode ocr |
| 表格/数据截图转 Markdown 表格 | vision.py "<图>" --mode table |
| 代码/日志/报错截图 | vision.py "<图>" --mode code 或 --mode error |
| 小字看不清：先全图定位，再裁局部放大读 | vision.py "<图>" --crop x1,y1,x2,y2 --budget large |
| 多张图对比 / 批量读 | vision.py "a.png" --images "b.png" "c.png" --prompt "对比这两张" |
| 高分辨率细节（4K 截图小字） | vision.py "<图>" --budget large |
| 原图直发（不缩放） | vision.py "<图>" --no-resize |
| 环境自检（配置/PIL/接口） | vision.py --check |
高质量识图工作流
- 先整体：用默认参数读一遍，拿到全局描述并定位疑点（小字、报错、表格局部）。
- 再局部：对疑点区域用
--crop x1,y1,x2,y2裁出来，配合--budget large放大后再读，直到信息足够。 - 关键内容原样转述：报错码、数字、代码、日志必须逐字转述，不概括、不脑补。
--crop 坐标为原图像素坐标（左上角为原点）；--save-crop 路径 可把实际发送的裁切图存下来复核。
注意
- 不要假装看到了图片，必须先运行脚本拿到描述再回答。
- 脚本报错（文件不存在、超过大小限制、未配置、API 失败）时如实转述错误并给出建议。
- 多张图一起发送时，
--crop只作用于第一张主图。 - 脚本默认关闭模型思考（
thinking: disabled，识图更快）；模型不支持该参数时删掉对应字段，或改为adaptive开启。 - 输出乱码（旧版控制台代码页 936）时先执行
[Console]::OutputEncoding = [System.Text.Encoding]::UTF8; $env:PYTHONIOENCODING='utf-8'再跑脚本；脚本本身已对交互终端/管道分别处理编码。 - 脚本重定向到文件时产物是 UTF-8，读取用
Get-Content -Encoding UTF8，否则中文会花。
安装到具体宿主
- DSH 接入（含框架补丁步骤，人类安装者才需要）：references/dsh-integration.md
What ships with it
3 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.
- today Changed · -19 lines · +2 tokens per session 63337a30c577
- 4d ago First seen · 91 lines · 45 tokens per session scan A 38690ea1d1ac
vision-skill is a skill published in the GitHub repository DDDFXYqiming/Agent_Extensions (7 stars, last pushed yesterday), licensed MIT. It adds 47 tokens to every session and 1,140 once invoked, about $0.0002 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 2026-08-31.
Other skills, from other repositories
tidyfactor-styler
Production framework styler and surgical RTL UI polish engine with Contextual Decision Layer (CDL). Operates directly inside existing codebases across React/Next.js, PHP, WordPress, and Vanilla HTML/CSS/JS. Trigger on commands 'brief', 'component', 'section', 'page', 'redesign', 'typography', 'palette', 'layout'…
setup-matt-pocock-skills
为本仓库配置工程技能——设置其 issue tracker、分诊标签词汇表和领域文档布局。首次使用其他工程技能前运行一次。.
scaffold-exercises
创建包含 section、problem、solution 和 explainer 的练习目录结构，且能通过 lint 检查。当用户想要搭建练习框架、创建练习模板或设置新的课程章节时使用。.
migrate-to-shoehorn
将测试文件从 as 类型断言迁移到 @total-typescript/shoehorn。当用户提到 shoehorn、想要在测试中替换 as、或需要部分测试数据时使用。.
writing-for-agents
为智能体编写文档。当你正在创建或编辑技能，或者修改 AGENTS.md / CLAUDE.md 时使用。.
code-review
从固定点（commit、branch、tag 或 merge-base）开始，沿两条轴线审查变更——规范（代码是否遵循仓库文档化的编码规范？）和规格（代码是否与原始 issue/PRD 的要求一致？）。两条审查线在并行子 agent 中运行，并以并排方式报告结果。当用户想审查一个分支、PR、进行中的变更，或要求"从 X 开始审查"时使用。.