deepseek-eyes

Overview

deepseek-eyes：给 DeepSeek Harness 的纯文本模型装上眼睛——把粘贴的图片经 OpenAI 兼容视觉 API 转成文字描述，再转发给真实的文本模型（Give text-only models in DeepSeek Harness eyes: describe pasted images via an OpenAI-compatible vision API, then forward to the real provider. Not affiliated with DeepSeek.）

Community · Topic auto-analysis — Automatically discovered through the dsh-plugin GitHub Topic. Static checks verified a pinned public root bundle contract and committed files without executing third-party code. This is not a human review, security audit, runtime smoke test, publisher identity check, or official endorsement.

Install

Install this bundle from GitHub

The CLI validates the bundle, invokes native DSH, then reports completion.

npx dshpub add fryghost/deepseek-eyes --ref a0abfaad1401a3ee83ec53db272ed98cde3f2531

Capabilities

What it contributes

HostCordis loadableNot inspected

Client / UIWeb UINot inspected

Model toolsNot inspectedNot inspected

Profile stateconditionalNot inspected

README / EN

Package documentation

Give a text-only model in DeepSeek Harness sight — then just paste the image.

LIMITATIONS

Known limitations

- The text model receives a **description**, not pixels: fine-grained geometry, exact colors, and pixel-level layout are out of scope (use dsh-vision-toolkit alongside). - The delegated call is intentionally unmarked, so sessions on the bridge route lose adapter replay state for cached-response replay (correctness is unaffected; replayed responses fall back to a fresh provider call). - Only PNG / JPEG / WebP / GIF image parts admitted by the DSH attachment path are handled; the vision endpoint must accept the chosen formats. - Descriptions are generated per image block, sequentially, inside the request path: budget `vision.timeoutMs` accordingly. - The in-memory cache is process-local and clears on restart.