DeepSeek Harness Plugins · Model Inference
Plugins that extend DeepSeek Harness (DSH) with vision, workflows, memory, web tools and more.
Model-inference plugins sit between DeepSeek Harness and the model that answers. Some route a long job across a cheap drafter and a stronger checker. Some let a text-only model see a screenshot. Both kinds can change your bill and the host that receives the repository.
Open this category when the default model is the bottleneck: it cannot see an image, or it is the wrong price for every step of a long run. Do not open it because a comparison table was inconclusive. Deciding after a comparison settles that with a prompt, not with a router.
Read the routing table before you point the plugin at private code. Write down which model handles which step. Try it on a public sample. A vision adapter should be tested on a screenshot whose label you can read yourself. If it names the wrong button, it is not ready for a UI review. Editor's notes on ModLens and the routing suite, when you are on those pages, spell out that first test. Other cards in this category do not inherit those notes.
This shelf overlaps client plugins that merely display a balance. A balance widget does not choose the model. A router does. Installing both on the same afternoon means you will not know which one spent the account. The category guide draws that line.
Uninstall belongs in the README before the install command does. An inference plugin that fails on startup can take the session with it. Have the removal steps open. Stars measure attention, not compatibility with today's Harness build.
Log the model name on each step for one real task before you trust the route. If the log disagrees with the table you wrote, the table is fiction. Fix the config or remove the plugin. Do not add a second router to correct the first.
Install in dsh with no login or API key for unlimited free access to frontier models such as DeepSeek V4.1 Flash and Kimi K3, plus a local OpenAI-compatible gateway.
A unified suite combining hot runtime injection, task-aware thinking-mode routing, and a graded session protocol with red-team gates to sustain model diligence across long-horizon inference.
ModLens is a vision plugin for DeepSeek Harness that gives text-only models sight by reading images pasted directly into chat, with zero-config setup and multiple vision engines.
Improves token efficiency in DSH via low-intrusion OPS tools, selective hiding of official tools, and on-demand dynamic loading for Native and PTC modes.
On-demand vision for text-only DeepSeek Harness agents: built-in free keyless vision chain and 14 vision tools, routing image turns as tool calls to vision models with pixel fidelity, no Python needed, one-command install.
Give text-only models in DeepSeek Harness eyes, enabling image Q&A, long-screenshot OCR, UI restoration, and GUI visual tasks.
An armor-breaking plugin for DeepSeek: rewrites instruction priority, expands refusal-blocked phrases, persists across sessions, and shows a green active indicator.
Provides workspace @ path references for the DeepSeek Harness web interface, with searchable file/directory picker, filter rules, and path handling.
A DeepSeek Harness settings plugin for prompt jailbreak, multi-rule management, permission and tool patching with auto-reapply after npm upgrades, plus Windows CMD silence fixes.
Use your ChatGPT (Codex), Claude, Grok (X Premium) and GitHub Copilot subscriptions as LLM providers in DeepSeek Harness — no API keys, with OAuth login, usage stats, and image/video generation tools.
Bring native conversational image generation and editing to DeepSeek Harness, supporting text-to-image, image-to-image, multi-turn editing, and multiple image providers.
A cost-aware model-routing workbench for DeepSeek Harness Desktop to plan tasks locally, compare routes, inspect execution permissions, and run dependent work packages via model APIs or official vendor CLIs.