deepseek-vl-support
Give DeepSeek (text-only) models vision in Claude Code, Codex, and Agent Plugins clients: describe images via any OpenAI-compatible vision endpoint.
790 results
Give DeepSeek (text-only) models vision in Claude Code, Codex, and Agent Plugins clients: describe images via any OpenAI-compatible vision endpoint.
A DeepSeek-style paperclip attach button in the DSH web composer; uploads files to the DSH host (model-aware: native image blocks for multimodal models, file paths for text-only models).
DeepSeek Harness plugin that connects to a PowerContext Server over HTTP for recall, memory, handoff, experience, and skills.
Reduce large agent tool output by what it means, not by where it was cut.
Headroom-inspired automatic context compression plugin for DeepSeek Harness (dsh): compresses tool outputs before they reach the model and keeps every lossy compression reversible via CCR retrieval tools.
OpenViking retrieval, resource management, auto-recall and session memory for DeepSeek Harness.
Make your AI harness speak — voice announcements for DSH and other AI coding harnesses (Windows SAPI5 + macOS system voices)
Voice-first session loop for DeepSeek Harness: a composer microphone button with browser/local speech-to-text (Web Speech, FunASR, whisper.cpp), a speak tool for text-to-speech replies (browser, edge-tts, piper), event announcements with mute, and speak-to-interrupt.
Persistent project memory for dsh agents: index docs (PDF/Markdown/text) and code symbols into a searchable per-workspace store, recall them with cited sources, and keep experience entries (problems -> solutions) searchable on demand.
dsh plugin: recognize attached images with the built-in Windows OCR engine (Windows.Media.Ocr) and send only the recognized text to the model. Text models never receive image bytes; vision passthrough is opt-in.
PDF toolbox for DeepSeek Harness: extract text, metadata, and page ranges from PDF files (local parsing, no API key).
会话级自动上下文压缩插件:回合中每步前 + 回合结束检测上下文用量,超过会话阈值自动 compact,摘要注入上下文后自然继续(DeepSeek Harness / DSH)
AI 输入框粘贴大段文本自动缩略为 [Paste #N +M lines] 小序号标签(编号不可改);👁 点击预览全文(点外关闭)、✎ 编辑名称、↺ 恢复纯原文;无删除按钮,删输入框标签文本即消失;发送时自动展开原文并加固定方括号纯标识块分隔(编号与输入框一致);slash 命令参数展开不加装饰;代理可用 myds_read_captured 工具按编号读取标签对应文本。
DSH peer link — point-to-point messaging between dsh agents and other local agent sessions (e.g. Claude Code) over unix sockets. Independent plugin: register as a peer, receive messages into agent context, reply with peer_send, list peers with peer_list.
Session folders for the DSH web sidebar: one-level named folders per workspace to group sessions, drag-and-drop or context menu to move them, server-side persistence. Status badges mirror the built-in session browser.
Voice input plugin for DeepSeek Harness: microphone → local/browser speech recognition → text submitted as a normal chat message (input-only, preset-agnostic)
Eyes for text-only DeepSeek on DeepSeek Harness: image understanding via Doubao Web by default (zero cost, no API key), Antigravity IDE quota (flash/pro), Gemini API, or Cockpit proxy — model-invokable vision tool, wrapper adapters, evidence memory, and a bilingual client panel.
A searchable, sortable model picker for the DeepSeek Harness Web UI — provider groups, favorites, live per-task cost, prices and context windows.
Persistent long-term persona memory for DeepSeek Harness: MEMORY.md/USER.md storage, memory + memory_search tools, per-request context injection, secret scanning, and a WebUI memory-management settings page. Ported in spirit from pi-hermes-memory. TypeScript sources, built with tsc + esbuild.
Real-time token consumption HUD for DeepSeek Harness: a sleek tech-style panel showing live token usage, context pressure and cost estimates across sessions
Rewind a DSH conversation to any user message: hide the tail from the chat view and the model context, keep the durable log intact, cancel freely before sending.
CodeGraph for DSH — registers native `codegraph_*` tools (core surface: status/init/sync/explore; full surface adds query/node/files/callers/callees/impact/affected/index/uninit) wrapping the `codegraph` CLI, plus a prompt front-load hook that steers structural CodeGraph context into structural prompts, so the DSH agent can bootstrap, maintain, and query a pre-indexed code knowledge graph without grepping files.
dsh-skillport — every skill you already have (Claude Code, Codex, Cursor, Gemini CLI) works in DSH: Agent Skills SKILL.md discovery across .claude/.agents/.dsh/.gemini + extraPaths, Tier-2 conversions (cursor rules, claude commands, context files), find_skill search, and a skills doctor
Third-party DSH WebUI enhancement plugin: custom backgrounds, theme colors, prompt presets, token visibility, and manual context compaction.