DSH Vision Toolkit
DeepSeek Harness-native vision bundle that gives text-only agents structured tools for image Q&A, OCR, grounding, UI restoration, pixel diffs, and artifacts.
Install
$ **Install:** dsh plugin --profile web add @anionex/dsh-vision-toolkitRepository Intelligence
AI-assisted, source-grounded explanation based on the public repository snapshot. It does not replace compatibility or security verification.
Key Capabilities
- Provides image Q&A, image grounding, element detection, and long-screenshot OCR through a configured OpenAI-compatible vision endpoint.
- Offers local image operations including crop, tracing to SVG, pixel-diff analysis, foreground extraction, dominant-color analysis, and HTML screenshot rendering.
- Uses a versioned vision-tools Skill and agent-scoped progressive tool exposure.
- Creates structured artifact outputs such as PNG, JPEG, SVG, JSON, Markdown, and OCR-related files.
- Integrates with DSH Credentials, Settings, Session logs, and Web artifact presentation.
- Supports installation into DeepSeek Harness Web and Headless profiles.
Useful For
- Answer focused questions about workspace images.
- Extract OCR from long screenshots.
- Ground or detect UI elements using original-image pixel coordinates.
- Restore a screenshot or sketch into editable HTML/CSS and measure visual differences.
- Compare reference and generated screenshots with pixel-diff reports.
- Crop, trace, analyze, or extract foregrounds from local images.
Who It Fits
- DeepSeek Harness users working with text-only agents on visual tasks.
- Developers performing screenshot-based UI restoration or regression checks.
- Agents and workflows that need structured visual artifacts and coordinate-based results.
Documented Limitations
- Requires Python 3.11 or newer for the runtime.
- Remote image Q&A, grounding, detection, and non-split-only long-screenshot OCR require an OpenAI-compatible vision endpoint and DSH Credential.
- Managed runtime preparation may require network access unless required packages are available in the configured cache.
- vision_html_screenshot requires Chrome, Chromium, or Edge.
- Input images are limited to PNG, JPEG, GIF, or WebP files within the session workspace or configured allowed directories.
- Current scope excludes Web upload, camera/video/audio/document ingestion, interactive box editing, automatic GUI clicking, model routing, and cross-session vision caches.
DSH Compatibility
Version-specific runtime evidence collected by DSH Plugin. A missing result means we have not tested that combination yet.
Security Signals
Objective signals discovered from package metadata and source inspection. These are not a guarantee that a plugin is safe.
package.json declares 1 runtime, 13 development, 21 peer, and 0 optional dependencies.
package.json declares DSH bundle metadata.
A root package.json was captured and can be inspected by the registry.
GitHub reports the repository license as MIT.
Public GitHub source metadata is available for this registry snapshot.
package.json declares 1 runtime, 13 development, 21 peer, and 0 optional dependencies.
package.json declares DSH bundle metadata.
A root package.json was captured and can be inspected by the registry.
GitHub reports the repository license as MIT.
Public GitHub source metadata is available for this registry snapshot.
package.json declares 1 runtime, 13 development, 21 peer, and 0 optional dependencies.
package.json declares DSH bundle metadata.
A root package.json was captured and can be inspected by the registry.
GitHub reports the repository license as MIT.
Public GitHub source metadata is available for this registry snapshot.
package.json declares 1 runtime, 6 development, 20 peer, and 0 optional dependencies.
package.json declares DSH bundle metadata.
A root package.json was captured and can be inspected by the registry.
GitHub reports the repository license as MIT.
Public GitHub source metadata is available for this registry snapshot.
package.json declares 1 runtime, 6 development, 20 peer, and 0 optional dependencies.
package.json declares DSH bundle metadata.
A root package.json was captured and can be inspected by the registry.
GitHub reports the repository license as MIT.
Source & Registry Notes
Traceable source and registry metadata for this entry, kept separate from runtime verification.
- Source repository
- Anionex/dsh-vision-toolkit
- Registry source
- Public GitHub repository
- Source snapshot
- 4d79ae020f34
- Artifact type
- plugin-bundle
- AI enrichment
- gpt-5.6-terra · 2026-08-14
- Prompt version
- dsh-plugin-enrichment-v3
This project is independently indexed from public source information. DSH Plugin is not affiliated with DeepSeek or the plugin author. Always check the author repository before installation.
Repository Activity
- GitHub stars
- 332
- Forks
- 18
- Open issues
- 0
- Last commit
- 2026-08-14
- Last release
- 2026-08-14
Related DSH Plugins
Ranked by overlapping capabilities, use cases, plugin type, categories, and DSH profile.
DeepSeek Harness vision-routing plugin with provider fallback chains and pixel-level image tools.
View plugin →A DeepSeek Harness vision plugin that turns pasted images into structured OCR, layout, and semantic evidence for supported text-only models.
View plugin →An installable DeepSeek Harness bundle for robotics and embodied-intelligence research workflows.
View plugin →Adds an OpenAI-compatible vision bridge to DeepSeek Harness, forwarding images to a configured vision model and displaying descriptions in a web panel.
View plugin →