DeepSeek Harness Plugins
Plugins that extend DeepSeek Harness (DSH) with vision, workflows, memory, web tools and more.
Parses bilibili video links, extracts subtitles or transcribes audio via local Whisper, and exports SRT/TXT/JSON as an auto-invoked model tool.
Cursor-style element picker for DSH sidebar: hover, click, and send structured DOM references to your Agent for precise front-end edits on local HTML or web apps.
Add a Workspace button to the DSH Web UI sidebar so you can create or switch to a workspace by entering an existing directory path.
An out-of-tree PDF reader for DeepSeek Harness that unifies reading, annotation, multimodal AI analysis, and multi-document Q&A in one web workflow with optional OCR.
Per-provider and per-model LLM rate limiting for DeepSeek Harness with token bucket, concurrency control, upstream 429 backoff, and gateway identity rules.
Block, replace, inject, and reorder system-prompt sections and filter tool catalogs per session phase and agent preset, all from a settings panel, applied live.
A sidebar usage monitor for DSH Web showing live quota bars and per-model request budget estimates for OpenCode Go and Command Code, with bilingual UI and no extra process.
A Git status, commit, and push panel for the DeepSeek Harness web interface, with session-aware branch tracking and AI-generated commit messages.
Generates diverse candidates via multi-temperature sampling, then uses a judge call to select and return the best answer with reasoning.
Enhanced multi-tab side chat with whole-message quoting, zero-wait mid-stream fork, dual-mode dock coexisting with better-sidebar, and top-layer overlay.
Zero-dependency multilingual subtitle translation workflow for dsh: SRT/VTT parsing, LLM sentence translation, bilingual merge, alignment validation, and batch processing as dsh tools or CLI.
Aggregates nine search APIs into one prioritized fallback queue with multi-key rotation, per-attempt timeouts, and transparent error reporting for reliable web search in DSH.