DeepSeek Harness Plugins · Model Inference
Plugins that extend DeepSeek Harness (DSH) with vision, workflows, memory, web tools and more.
Packages 14 SuperPowers engineering skills, auto bootstrap, sandbox compatibility, and optional UI/UX design tools for DeepSeek Harness.
Prices the current session on the actual route and every alternative in the table, with time-of-use support and a cross-route cost comparison.
Adds a model extension settings page to DeepSeek Harness for configuring reasoning effort, input modality, and compatibility, with strict host version validation.
Automatically renders data-heavy answers as streaming narrative text, AntV charts, or a natural combination, with live theme adaptation and semantic data expression.
Use a local MLX-LM or MLX-VLM model as a DeepSeek Harness inference provider with optional DSH-managed server lifecycle on Apple-silicon macOS.
Connect LawStar legal data MCP in one click - paste your API Key to search statutes, retrieve clauses, identify citations, and verify AI hallucinations.
A DeepSeek Harness plugin bundle that adds the prompt-enhancer tool to rewrite vague instructions into structured prompts with role, task, constraints, and output format.
Adds an independent retry policy for PI_AI_ERROR fallback codes in DSH model inference, auto-retrying once after a 20s delay without interfering with the default exponential backoff.
Solves olympiad math (IMO/Putnam/USAMO) via pure reasoning, adversarial subagent verification, and calibrated confidence with honest abstention; optional LaTeX PDF output.
Local organic reaction prediction for DeepSeek Harness with retrosynthesis, forward prediction, route planning, and synthesizability scoring, powered by SynOmega.
Inject a custom prompt at the very top of every session's system prompt, editable live from the web settings page with instant effect on the next request.
Connects DeepSeek Harness to a logged-in Grok Build account with dynamic model discovery, streaming inference, image input, and optional Web/X Search.