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/gou-gee/deepseek-vision/visionnpx skills add GOU-GEE/deepseek-vision --skill visiongit clone --depth 1 https://github.com/GOU-GEE/deepseek-visionWrote 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/gou-gee/deepseek-vision/vision)<a href="https://agentmods.dev/skills/gou-gee/deepseek-vision/vision"><img src="https://agentmods.dev/badge/skills/gou-gee/deepseek-vision/vision.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 5.1 | $0.00092 | $0.02028 |
| Opus 5 | $0.00046 | $0.01014 |
| Sonnet 5 | $0.00018 | $0.00406 |
| Haiku 4.5 | $0.00009 | $0.00203 |
Grade A, and why
vision 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 5d 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 — 146 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Vision Skill — 图片理解
DeepSeek 主模型是纯文本模型,无法直接“看”图片。本 Skill 通过 MCP 工具
(analyze_image / analyze_clipboard / compare_images)把图片交给第三方
OpenAI 兼容视觉模型(如智谱 GLM-4.6V、硅基流动 Qwen2.5-VL、通义千问
qwen-vl-plus),再把识别文本返回给主模型。
触发条件
满足以下任一条件时,必须调用视觉工具,而不是假装能看到图片:
- 用户输入中包含本地图片路径,例如
./screenshot.png、~/Desktop/photo.jpg、 或绝对路径/Users/me/1.png。 - 用户输入中包含图片 URL,例如
https://example.com/foo.png。 - 用户输入中包含 base64 图片数据(可能带
data:image/...;base64,前缀)。 - 用户提到剪贴板:说「看看剪贴板里有什么」「我截图了」「刚复制的图片」
——调用
analyze_clipboard。 - 用户要求对比多张图片:说「对比这两张图」「这两张有什么区别」——调用
compare_images,把各图片路径/URL 放进images数组。 - 用户明确要求识别 / 理解 / 描述图片内容,例如:
- “帮我看看这张图里有什么错误”
- “提取这张图片里的文字”
- “这张截图显示了什么报错信息?”
- “描述一下这张照片的场景”
- 用户上传了图片但主模型无法直接处理(例如在 DeepSeek Harness / Codex 中以附件形式传入),此时同样应调用工具。
必须执行的操作
-
先确认工具可用:调用前确认当前会话工具列表里有对应的视觉工具。 若没有,说明 MCP 未加载,提示用户检查配置并重启,不要硬编造结果。
-
判断图片输入形式:本地路径 / URL / base64 / 剪贴板,选择对应工具:
- 单张图片 →
analyze_image(image=..., prompt=?, task=?) - 剪贴板图片 →
analyze_clipboard(prompt=?, task=?) - 多图对比 →
compare_images(images=[...], prompt=?)
- 单张图片 →
-
选择
task(预置任务,无需自己拼 prompt;也可用prompt自定义指令,prompt优先级更高):task 用途 效果 describe(默认)通用描述 详细描述图片内容 ocr提取文字 逐字提取,保留排版换行,不做评论 describe_uiUI 截图 布局 / 组件 / 可见文字 / 状态 diagnose_error报错截图 错误信息 / 原因 / 修复步骤 understand_diagram流程图/架构图 类型 / 组成 / 关系 / 目的 analyze_chart数据图表 类型 / 坐标轴 / 趋势 / 洞察 code_from_screenshot代码截图 语言 / 可编辑代码块 自定义
prompt示例:- 提取文字:
请提取图片中的所有文字,保持原有排版顺序。 - 找错误:
请识别图片中的错误信息,并说明可能的原因。 - 描述场景:
请描述图片中的场景、人物、物体和环境细节。 - 其他需求:把用户的原始诉求改写为对视觉模型的清晰指令。
- 提取文字:
-
调用 MCP 工具:
analyze_image(image="<路径或URL或base64>", task="ocr") analyze_clipboard(task="diagnose_error") compare_images(images=["./a.png", "https://example.com/b.jpg"], prompt="对比两者差异") -
工具返回 JSON:
{"success": true/false, "result": "...", "model": "...", "provider": "...", "fallback_used": true/false, "attempts": 0, "usage": {...}, "cached": true/false}。success: true:把result中的识别文本直接作为回答呈现给用户, 可适当补充说明(来自哪个模型);cached: true只表示命中同会话缓存, 内容仍可正常使用。success: false:把error字段(如IMAGE_LOAD_FAILED、VISION_API_ERROR、CONFIG_ERROR、CLIPBOARD_ERROR)和result中的 错误信息转告用户,并给出修复建议(检查路径/URL 是否有效、API Key 是否 配置、图片是否超限、剪贴板是否有图等)。
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.
- 5d ago First seen · 146 lines · 92 tokens per session scan A a2d2cb0724c1
vision is a skill published in the GitHub repository GOU-GEE/deepseek-vision (5 stars, last pushed 10d ago), licensed MIT. It adds 92 tokens to every session and 2,028 once invoked, about $0.0005 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
submit-dsh-plugin
验证并提交 DeepSeek Harness 插件到 imsai-sh/awesome-deepseek-harness-plugins 社区目录。适用于插件作者要求收录、发布或提交自己的插件,创建目录 JSON,修复目录提交 PR,或者发起合规 PR。检查公开仓库、dsh-plugin topic、dsh.bundle.patch、作者测试证据、双语元数据和单文件差异,并在获得授权后执行 fork、push 和创建 PR。.
dsh-plugin-guide
Use when developing, reviewing, packaging, debugging, or answering questions about DeepSeek Harness (DSH) plugins — the plugin-based agent harness on vendored Cordis. Applies the official plugin-development constraints (plugin contract, cordis.yml layers, services/events/effects, tool DSL, bundles/profiles) backed by…
ito-compute
Query live GPU inventory, submit an authenticated Itô fixed-rate RFQ, inspect RFQ or procurement status, revoke device credentials, and run explicitly gated node qualification through the separately installed canonical CLI. Use when a user asks to find H100/H200 capacity, request a fixed compute rate, check Itô…
ito-inference
Inspect the availability of model serving on a completed Itô compute booking and, when the canonical backend becomes available, hand off an explicitly confirmed serving manifest. Use after ito-compute has booked GPU nodes and the user asks for an OpenAI-compatible endpoint, ito-serve, hosted Kimi, or self-hosted…
ito-training
Inspect the availability of ML training on a completed Itô compute booking and, when the canonical backend becomes available, hand off an explicitly confirmed training manifest. Use after ito-compute has booked GPU nodes and the user wants pre-training, fine-tuning, or RL on that metal. ECC implements no training…
dsh-plugin-schemes
开发任何 deepseek-harness 插件前,先按本规范选对形态,再动手。.