dsh-vision-sidecar

Add no-key hosted or OpenAI-compatible vision to DeepSeek Harness while keeping the configured reasoning model 适合已有文本模型却缺图像能力的场景；需配置 vision 端点与目标模型，按请求约束转发。

パッケージ

dsh-vision-sidecar

互換性

未検証

Harness ピア範囲

>=0.1.0-rc.6 <0.2.0

Cordis ピア範囲

>=4.0.1 <5

バージョン

0.1.4

ライセンス

MIT

最終更新

2026/08/14

Configuration

The no-key LLM7.io default needs no patch. To change the reasoning target or request bounds, override the vision-sidecar row:

- id: vision-sidecar
  config:
    targetProvider: your-existing-text-provider
    targetModel: your-existing-text-model
    visionBaseURL: https://api.llm7.io/v1
    visionModel: default
    visionApiKeyEnv: ''
    visionTimeoutMs: 60000
    visionMaxResponseBytes: 524288
    visionMaxSessionBytes: 1048576
    maxImagesPerRequest: 4

Remote URLs must use HTTPS. HTTP is accepted only for loopback-compatible development endpoints. URL-embedded credentials, query strings, and fragments are rejected.

dsh-vision-sidecar

インストール

ドキュメント

Configuration