232 results
Structured context for DeepSeek Harness: split the conversation into typed sub-contexts (reasoning / tool calls / replies) and index them with a table of contents. 结构化 dsh 上下文:按类型拆分小上下文(思考/工具调用/正式回复),以结构化目录为索引。
Browser patrol and website inspection plugin for DeepSeek Harness: teach once, replay deterministic runbooks with managed Chromium, credentials, screenshots, checkpoints, and selector recovery.
Windowed, reversible context management for DeepSeek Harness; replaces Basic compaction.
Turn successful agent tool calls into reviewed, parameterized routines you can actually replay. DSH, MCP, and WebMCP.
OpenViking memory & context bundle for DeepSeek Harness: auto-recall, session capture, threshold commit, offline outbox replay, self-contained MCP tool bridge, viking:// URI guard.
Hash-anchored read/replace/undo tools for DSH. Every line of a file carries a unique 3-char content hash; edits target hashes, never line numbers, so stale anchors are caught before they touch the file. Port of pi-hashline-edit-pro for the Pi coding agent.
DeepSeek Harness bundle: a Chrome browser for agents (browser_goto / browser_evaluate / browser_screenshot and 12 more) plus a no-key net search tier (browser_search / browser_fetch / Reddit tools, ported from pi-lynx) with a live two-way pane inside the DSH Web GUI. The browser engine is a fork of zenbu-labs/terminal-browser with the React Ink terminal UI replaced by a DSH tool surface and web pane.
Evidence-governed experience memory for DeepSeek Harness
Register each agent preset's own skills/ directory as discoverable, preset-isolated skills in DeepSeek Harness, including blank-session preset switching and a startup roster prewarm that shrinks the web catalog warm race. Replacement for the old preset-skills copy-in mechanism.
Enforced pain-point-check guard plugin for DeepSeek Harness: after two non-converged experiments on one problem it injects the three questions, denies non-investigative tool calls until they are answered in the reply text, and blocks same-direction retries
Agentic Surface Compaction (ASC) for DeepSeek Harness: the model decides when and what to compact, committed as durable session-log replacements with full replay, search, and degradation
System-level Web Notification when the model finishes a reply in the DeepSeek Harness web GUI
Skill usage statistics for DSH: counts every skill invocation across sessions (replay of existing logs + live session/event subscription), detects zombie skills, and helps decide which skills to keep.
Edit a sent user message and resend it: fork a new branch from that message's turn and regenerate the reply with the edited text.
DeepSeek Harness plugin: a third-party image-generation relay. Adds the 生图配置 card (Settings → Plugins → 插件配置) for configuring an OpenAI-compatible (/v1/images/generations + /v1/images/edits), DashScope (通义万相 async-task), Replicate or fal.ai image API; registers image_generate (文生图/图生图/局部重绘/指令编辑), imagegen_status, imagegen_test, imagegen_configure tools; installs a 生图 skill so the model auto-triggers on 生图/画图 mentions.
Message Annotator for DeepSeek Harness: stage numbered quote+comment annotations on assistant messages (selection floating trigger + per-message action), then send them aligned and together to the agent
Text-to-speech for DeepSeek Harness: speak agent replies in the Web UI with a provider fallback chain (OpenAI, ElevenLabs, Google, Azure, Groq, Deepgram, OpenRouter, Edge, Piper, eSpeak).
DSH web plugin: replace the DeepSeek Harness branding (tab favicon, sidebar wide logo, collapsed rail icon) with your own images — upload two pictures from the settings dialog, no manual resizing
Controlled A/B comparisons and evidence-backed reports for DeepSeek Harness plugins and presets.
Replace DSH web tool icons with emoji: think 🤔, read/search 🧐, pwsh 👨💻, edit ☝️🤓, error 😫
Shadow-replaces the built-in glob tool: auto-excludes bottomless directories (node_modules, dist, build, .venv, ...) and requires an explicit include whitelist to search inside them.
Keeps Charm Hyper requests under the gateway's 10 MiB body cap: replaces the oldest images with placeholders (newest survive) before sending, and retries with a tighter projection when the gateway still rejects the body.
Voice conversation mode for the DeepSeek Harness Web GUI: mic input with auto-submit, streaming spoken readout of replies, and a hands-free multi-turn loop with barge-in.
DSH 桌面消息提醒插件：模型需要你操作（审批/方案确认/提问的可操作提醒）或回复在后台完成时，系统通知+提示音+标签页标题提醒你；可在设置页开关并切换通知语言。Desktop notification & message alerts for DeepSeek Harness web — toasts, sounds and a tab-title marker when the agent needs your input (approval / plan review / question) or a reply finishes in the background; toggleable from Settings with zh/en notification language.