dsh-visual-aid
DSH visual-assistance bundle that lets text-only main models work with image-containing chats through a configured vision model.
Install
$ dsh plugin --profile web add @sy008/[email protected]Plugin Overview & Capabilities
AI-assisted organization based on the public repository snapshot. The content must be grounded in source evidence and does not replace compatibility or security verification.
Key Capabilities
- Detects image blocks in conversations and generates descriptions with a configured vision model
- Replaces images in main-model requests with numbered text placeholders
- Provides view_image for follow-up questions about an original numbered image
- Provides visual_read_image to read image files through the plugin's own path
- Adds per-session enablement through /visual-aid on|off and a Web UI control
- Adds Visual Aid settings and a visual Q&A panel with history, discard warnings, and token statistics
- Persists per-session image descriptions and Q&A history, and inherits them when sessions are forked
- Maintains monotonically increasing image numbers after compaction
- Includes an optional multimodal-masquerade setting for model-switch checks
Useful For
- Using a text-only main model in a DSH conversation that includes screenshots or other images
- Letting a main model ask targeted follow-up questions about previously uploaded images
- Avoiding repeated image-description calls after branching from a session with existing visual data
- Reviewing visual-model interaction history and token statistics in the DSH Web interface
Who It Fits
- DeepSeek Harness users running text-only main models
- Users who can configure an image-capable vision-model provider and model
- DSH Web users who need image assistance without modifying DSH core
Documented Limitations
- Requires a configured vision model that supports image input; when none is configured, the plugin treats itself as disabled
- Image descriptions and visual Q&A consume vision-model calls and tokens
- DSH's normal model-switch protection may still reject switching an image-history session to a text-only model; the README recommends compaction or a new session
- Multimodal masquerade is documented as a risky option and can still result in UNSUPPORTED_CONTENT if Visual Aid is not functioning
- Visual data is stored separately per session under the user's .dsh/visual-aid directory; deleting those files loses generated descriptions and Q&A history
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.
GitHub reports the repository license as MIT.
No root package.json was captured in the latest GitHub snapshot.
Public GitHub source metadata is available for this registry snapshot.
Source & Registry Notes
Public provenance, Registry classification, and the latest source check for this entry, kept separate from runtime verification.
- Source repository
- shenyan-008/dsh-visual-aid
- Registry source
- GitHub · dsh-plugin topic
- Registry classification
- Plugin Bundle
- Source checked
- 947c22f · 2026-09-13
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
- 1GitHub stars
- Forks
- 0
- Open issues
- 0
- Last commit
- 2026-08-19
- Last release
- No release detected
Related DSH Plugins
Ranked by overlapping capabilities, use cases, plugin type, categories, and DSH profile.
Preview, create, and edit spreadsheets, docs, slides, and canvases in DeepSeek Harness powered by Univer.
View plugin →Compatibility bridge and host ABI enabling unmodified Pi plugins to run natively on DeepSeek Harness.
View plugin →A bilingual DeepSeek Harness cost-tracking plugin with session and daily costs, budgets, pricing management, balance lookup, and coding-plan quota displays.
View plugin →Persistent multi-model Agent teams and bounded DAG workflows with GUI controls for DeepSeek Harness.
View plugin →