dsh-plugin-effort-declare
Run the following command in DeepSeek Harness:
dsh plugin install zimodzh/dsh-plugin-effort-declare
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install zimodzh/dsh-plugin-effort-declare in your terminal to install this plugin, then restart DSH and open the Reasoning Tiers page in Settings; full source and docs are at https://github.com/zimodzh/dsh-plugin-effort-declare
About this plugin
Third-party OpenAI-compatible models manually added in DeepSeek Harness do not declare reasoningEfforts or input by default, so the chat interface shows no Effort picker and rejects image attachments even when the model fully supports them.
dsh-plugin-effort-declare adds a Reasoning Tiers page to Settings, letting you toggle available reasoning levels per model, declare image input, and write into the official llm-pi-ai namespace. It supports rewriting on-the-wire names to match gateway-specific spellings, ships presets for DeepSeek-compatible and OpenAI-compatible dialects, and never touches API keys, endpoints, or model lists.
Built for users who have manually added self-hosted or third-party OpenAI-compatible providers in DeepSeek Harness and want the Effort selector and image attachment to work. Requires DSH 0.1.2-alpha.2 or later.
Use Cases
- Manually added self-hosted models like Ollama or vLLM in DSH need an Effort selector to appear
- Third-party vision models are already configured but cannot upload images for multimodal chat
- Different gateways use inconsistent tier names and need on-the-wire renaming to pass parameters correctly
Best For
- Users who manually add third-party OpenAI-compatible models in DeepSeek Harness
- Developers needing per-model granularity over reasoning tier and image input declarations
- Power users managing self-hosted inference gateways (Ollama, vLLM, etc.) through DSH
Related Plugins
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.
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.