deepseek-vision-bridge
Add Vision and Image Generation to DeepSeek
Text-only DeepSeek clients cannot interpret pasted images or produce image files. This skill deploys a proxy using OCR, vision models, and generation backends.
請勿自動安裝此 Skill。
權威政策要求在執行任何安裝操作前由操作人員審查。
使用我的智慧代理安裝
將此請求複製給你的智慧代理,其中包含權威 Skill 頁面和清單。
Review the Skillstore skill "deepseek-vision-bridge" from https://skillstore.io/skills/lightlw-deepseek-vision-bridge.md and its manifest at https://skillstore.io/api/skills/lightlw-deepseek-vision-bridge/manifest. Verify the artifact. Do not auto-install. Inspect the skill and report your findings, then wait for an operator or manual installation decision.你的智慧代理仍應顯示其計畫,並依安全政策請求必要確認。
Agent 可讀資源
當 AI Agent、爬蟲或腳本需要乾淨脈絡、而不是讀取完整頁面時,請使用這些連結。
測試它
正在使用「deepseek-vision-bridge」。 Check whether this computer can run the bridge.
預期結果:
Node.js is available. Ollama is not running. The proxy port is free. Cloud vision needs a configured SiliconFlow key.
正在使用「deepseek-vision-bridge」。 Set up local OCR and vision without cloud services.
預期結果:
The plan uses Tesseract and Ollama, keeps images local, and lists the configuration changes requiring approval.
正在使用「deepseek-vision-bridge」。 Why are pasted images not reaching DeepSeek correctly?
預期結果:
The proxy is not listening on the configured port. Start it, confirm the Codex base URL, then inspect the proxy log.
安全審計
嚴重Many static alerts are false positives caused by Markdown, templates, loopback addresses, and JavaScript template literals. Confirmed risks include shell injection, credential-bearing network forwarding, sensitive configuration access, unauthenticated all-interface listening, SSRF, and Windows autostart persistence. The generated tool output also contains an instruction that discourages image verification.
已確認的安全問題 (17)
顯示全部 17 條已確認問題
能力複核項目 (79)
這些是真實的本機能力,對此技能而言可能是預期行為,因此需要複核,但不會按已確認的惡意行為計分。
風險因素
📁 檔案系統存取 (50)
🌐 網路存取 (34)
🔑 環境變數 (42)
⚡ 包含腳本 (1)
⚙️ 外部命令 (50)
偵測到的模式
分享與引用此報告
分享版本化評估報告、中性徽章、嵌入卡片與引用。Skillstore 只報告證據,不判斷該 Skill 是否安全。
複製報告連結
https://skillstore.io/skills/lightlw-deepseek-vision-bridge/audits/1?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_reportMarkdown 徽章
[](https://skillstore.io/skills/lightlw-deepseek-vision-bridge?utm_source=security_passport_badge)HTML 徽章
<a href="https://skillstore.io/skills/lightlw-deepseek-vision-bridge?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/lightlw-deepseek-vision-bridge/security.svg" alt="Skillstore security assessment" loading="lazy"></a>嵌入卡片
<iframe src="https://skillstore.io/embed/skills/lightlw-deepseek-vision-bridge.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>學術引用 (APA · BibTeX · CFF)
APA 引用
lightlw. (2026). deepseek-vision-bridge security audit report (audit version 1) [Author version unspecified]. Skillstore. https://skillstore.io/skills/lightlw-deepseek-vision-bridge/audits/1BibTeX 引用
@techreport{lightlw-lightlw-deepseek-vision-bridge-2026,
author = {lightlw},
title = {deepseek-vision-bridge security audit report (audit version 1)},
institution = {Skillstore},
year = {2026},
number = {1},
url = {https://skillstore.io/skills/lightlw-deepseek-vision-bridge/audits/1},
note = {Author version unspecified}
}CITATION.cff
cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "deepseek-vision-bridge security audit report (audit version 1)"
version: "unspecified"
type: report
authors:
- name: "lightlw"
date-released: "2026-08-01"
url: "https://skillstore.io/skills/lightlw-deepseek-vision-bridge/audits/1"
identifiers:
- type: other
value: "skillstore:lightlw-deepseek-vision-bridge:audit:1"
description: "Skillstore immutable audit report identifier"
Skillstore 評分
評分依據 證據置信度: 中你可以打造什麼
Enable Vision for Codex
Route pasted images through OCR or a vision model before sending text to DeepSeek.
Build a Local Vision Workflow
Use Ollama and OCR to keep image analysis on the local machine.
Diagnose Proxy Failures
Check ports, credentials, model availability, logs, and upstream connectivity.
試試這些提示
Assess my system for DeepSeek Vision Bridge. Report missing requirements and do not change files.
Plan an OCR-only setup for Codex and DeepSeek. Keep images local and request approval before editing configuration.
Configure SiliconFlow vision with Ollama fallback. Explain data destinations, credential requirements, and each proposed file change.
Audit and harden the bridge. Bind loopback, add client authentication, restrict outbound URLs, remove shell execution, and verify behavior.
最佳實務
- Bind the proxy to loopback and require a separate client authentication token.
- Store API keys in a restricted local file and never print complete values.
- Review every configuration change and test OCR, cloud vision, and fallback routes separately.
避免
- Do not expose the proxy port to a local network or the public internet.
- Do not pass untrusted text through shell command construction.
- Do not enable cloud processing when images contain sensitive data.