186 results
DSH plugin: keep text-only models (deepseek-v4-flash / deepseek-v4-pro) as the session default, and automatically route requests that carry image content to a configured vision-capable model (deepseek-v4-flash-vision-exp) 鈥?no manual model switching (periscope).
DeepSeek Harness (dsh) plugin: route one request to a modality-capable fallback model instead of forcing the whole session onto a single model.
Persistent, searchable, per-project memory for the DeepSeek Harness: decisions, rules, and sprint context in a queryable DuckDB file, rules enforced on every model request, with a full management UI in the Web Client
Corner approval popup: answer pending permission requests from any session while the agent window is away
A GitHub-connected visual git panel for DeepSeek Harness — guided flow (branch → commit → push → PR → review → merge), commit graph with PR/CI badges, one-click GitHub login, and structured git tools for the agent.
DeepSeek Harness plugin: transcodes images to a base64-safe format (PNG) before a vision request, fixing gateways like LM Studio that reject non-PNG/JPEG/GIF data URLs with "'url' field must be a base64 encoded image."
Desktop notifications and a local chime for DSH main-thread completion and requests for attention
Preview and download the exact request frame (system prompt + tools + wire messages) that DeepSeek Harness sends to the model.
DeepSeek Harness plugin: keep every model request inside an allowed set per provider, redirect or refuse the rest, and warn when a request would make a llama.cpp router reload models
Turn approved tracker tickets into review-ready pull requests with DeepSeek Harness.
Review-first request-dialect preflight for custom DeepSeek Harness providers.
Authored PR feedback inbox with shared Codex/DSH progress, priorities and read-only GitHub collection.
Inspect raw LLM requests and responses in DeepSeek Harness Web
Keep DSH tool results cache-prefix safe: truncate a result before it first enters the session surface, and never rewrite a tool result a provider request has already delivered.
Request Context Profiler for DeepSeek Harness — see what changed between model requests, and how cache reuse changed with it.
DSH web plugin: better-sidebar Turn Review tab — a human gate on the diff of the just-finished agent turn (Approve / Request changes per path with comment). Not a rewind.
Use when turning a user idea or feature request into a DeepSeek Harness plugin; choose the right extension point and delivery shape, implement a Cordis plugin with correct lifecycle/effects/config, compose it, and run the required repository checks.
Approval timeout for DeepSeek Harness: when a sandbox/escalation approval is not answered within a configurable period, the request is automatically rejected instead of waiting forever.
IM bot notifications for dsh-TUI: push long-turn completion and approval requests to a 飞书/钉钉/企业微信/generic webhook, configured in /settings (webhook namespace), status via /webhook.
DeepSeek Harness plugin: content-routing LLM adapter. Text-only requests go to DeepSeek; requests containing images go to a user-configurable OpenAI-compatible vision provider. Read_image works natively for text-only models.
Automatic provider/model fallback chains for DeepSeek Harness agents when LLM requests keep failing (retry exhausted, auth, quota, rate limit)
Approve-for-me mode for DeepSeek Harness: an LLM safety reviewer auto-approves ordinary approval requests, while destructive high-risk commands still ask the human.
DeepSeek Harness plugin: Windows toast notification with sound when a session requests elevation approval — shows which session, why, and what is being elevated
DSH plugin: model reasoning levels plus CLI request mimic, with a unified Daily Optimization settings section.