DSH Vision Toolkit · AI Nav City

Core facts

Read by a script from the official API, not rewritten by a model.

Repository: https://github.com/Anionex/dsh-vision-toolkit
License: MIT
Languages: TypeScript
Maintainer: Anionex
Homepage: https://agent-vision.anionex.me
Verified: 2026-11-01

What it is

The DSH Vision Toolkit plugin provides 10 tools that can be called independently or composed into a workflow. The plugin keeps image understanding and deterministic local image processing in one Agent workflow. Its visual capabilities come from a packaged, pinned agent-vision-toolkit snapshot. The DSH plugin handles installation, session-scoped tool exposure, Credentials, path checks, cancellation, timeouts, result files, and Web presentation.

Main features

For routes that DSH positively identifies as text-only, the plugin registers a sibling (Vision Toolkit) variant.
By default, pasting an image in DSH Web switches to that variant and gives the model both a reusable workspace path and a visual description focused on the current task.
The bundled Skill tells the agent what to look at for different visual tasks, which tool to choose, how to proceed, and how to verify the result.
The bundled vision-skills Skill carries the complete upstream playbooks, explaining when to use each workflow, in what order to call the tools, and how to verify the result.
Coordinates always use original-image pixels in x1,y1,x2,y2 form, so grounding output can feed directly into cropping, tracing, or later automation.
The Web Settings panel exposes the full provider, runtime, timeout, image-limit, and image-input-variant configuration.

Getting started

The project says most users never need to configure the Python runtime: the plugin prefers a system Python 3.11+ and otherwise downloads a pinned standalone Python automatically from the domestic mirror, falling back to GitHub when the mirror is unreachable.
A normal installation does not require an agent-vision-toolkit source checkout or a local path setting.
With DSH Desktop, it bundles its own dsh CLI and intentionally does not add it to the system PATH.
Settings stores the Credential reference and never reads the saved secret back into the browser.
OpenAI Chat Completions-compatible endpoints, OpenAI Responses, and Anthropic Messages are supported.

From the README · sh

dsh plugin --profile web add @anionex/dsh-vision-toolkit

From the README · sh

dsh plugin --profile headless add @anionex/dsh-vision-toolkit

From the README · sh

dsh plugin --profile desktop add @anionex/dsh-vision-toolkit

From the README · yaml

- id: vision-toolkit
  config:
    provider:
      baseUrl: https://api.example.com/v1
      credential: MY_VISION_KEY
      model: your-vision-model
      protocol: openai

Who it is for

Native-image models and models with unknown input modalities are not wrapped. Do not put API keys, bearer tokens, cookies, or other secrets in provider.headers; continue to use provider.credential for the vision API key. Headers are injected only into URLs on the configured provider origin and beneath its base path.

Existing configurations continue to use Chat Completions. The Responses request sets store to false, but that flag does not replace reviewing the provider's own data-retention policy. Leaving reasoningEffort blank uses the model or proxy default, and higher reasoningEffort may consume more reasoning tokens and take longer.

The timeoutMs setting defaults to 30000 and accepts 1000 to 600000. Queue waiting and execution each receive the timeoutMs budget, so a queued call can take longer than timeoutMs overall. Logs distinguish queueMs from upstreamMs.

Configured shared roots are currently rejected on Windows because their ownership and access-control lists cannot yet be verified safely. Each inspection is a separate multimodal request containing the necessary intent and image, and context does not accumulate across calls. One update switched the default model to Gemini 3.7 Flash and fixed Qwen/Gemini bounding-box coordinate order.

Related tools listed on this site.