dsh-layered-memory
L0~L3 分层蒸馏记忆插件 for DeepSeek Harness:自动捕获对话(L0)、抽取原子记忆(L1)、整合场景块(L2)、蒸馏核心画像/团队方法论(L3),并在模型步骤前自动召回注入。移植自 MemoryCore (TencentDB Agent Memory) 的管线设计。
829 results
L0~L3 分层蒸馏记忆插件 for DeepSeek Harness:自动捕获对话(L0)、抽取原子记忆(L1)、整合场景块(L2)、蒸馏核心画像/团队方法论(L3),并在模型步骤前自动召回注入。移植自 MemoryCore (TencentDB Agent Memory) 的管线设计。
DeepSeek Harness plugin bundle: optimize raw instructions into professional Role / Task / Context / Format prompts through the harness llm service
VCC-style instant, near-lossless deterministic compaction engine for the DeepSeek Harness — a drop-in replacement for @deepseek-ai/dsh-compaction-basic
Command Code Go provider adapter for dsh: scan the Go-plan model catalog dynamically and serve /alpha/generate through the harness LLM seam
Configurable DSH-native LLM verifier with a Web settings page
Subagent Director: choose an LLM provider/model for each subagent (call args > role binding > plugin default > inherit parent) and plan main-agent/subagent responsibilities with role templates, plus a settings UI.
Per-conversation x-opencode-session header injection for OpenCode Go in DeepSeek Harness (DSH). Fixes 400 MissingSessionID, keeps prompt-cache routing optimal, scoped strictly to opencode.ai, runtime-switchable without restart.
One DeepSeek Harness Codex capability bundle for ChatGPT login, LLM access, Web Search, and durable Image Creation
Langfuse observability for DeepSeek Harness: OpenTelemetry traces, compaction, feedback Scores, and fork lineage
SGME 仓库级 DSH 插件包装(真实插件包在 adapters/dsh/sgme-bridge/,dsh-sgme 已发布 npm)。本文件让 dsh plugin add github:freehul/sgme 直接可装,并让 dshfind 能推导出 npm 安装命令。
DeepSeek Harness plugin: make a custom llm-pi-ai provider route masquerade as Claude Code / Codex — spoofed client-identity headers, plus an opt-in request-body fingerprint for relays that gate on the body.
鲸鱼电表:DeepSeek Harness 用量与费用面板 —— 人民币账单、错峰计价、可分享的 AI 账单卡片 | Token usage & cost dashboard for DeepSeek Harness (DSH), with peak/off-peak pricing and shareable bill cards
Local-model (Ollama) integration for DeepSeek Harness: discover, pull, remove, and inspect local models, route requests to them by task type or keyword with automatic fallback to the cloud, and get a one-shot status overview via /ollama.
Miraculous Standard — unified anchored agent preset for DeepSeek V4 Pro/Flash (official API & opencode-go). Minimal-exact two-tool first-request bootstrap, model-aware Pro/Flash paths, unified context gate, epoch-aware catalog management, rc7-ready shell handling.
Structural code intelligence for DeepSeek Harness (dsh) — gives the agent codegraph and codegraph_index tools to find where a symbol is declared, what calls it, and what a change reaches, from a tree-sitter index it builds itself. Compatible with the codegraph CLI's on-disk format.
Drop or paste images, PDFs, office docs, zips, videos and audio into the DSH composer as workspace file paths instead of model attachments — lets a text-only model agent read them via vision tools.
DeepSeek Harness plugin: paragraph-batched streaming reveals with 8 designed animations, smooth scroll-follow and a settings panel.
Per-model capability declaration for DeepSeek Harness: reasoning-effort levels (with wire spellings) and request modalities (text/image) for OpenAI-compatible providers — one settings section, no YAML hand-editing.
DSH prompt 工具箱:预制 + 自定义 prompt 模板,点击即插入当前对话输入框(v1 常规模式 + v1.1 智能模式悬浮卡) | Prompt toolbox for DeepSeek Harness: preset + custom prompt templates, one-click insert into the conversation.
Unified DeepSeek Harness plugin: role-based subagent routing + per-agent evolution — prefercmd/memory as knowledge allow/deny lists, so repeated tasks start from proven commands and save tokens
Reduce large agent tool output by what it means, not by where it was cut.
为 DeepSeek Harness 提供外挂识图模型:网页配置面板、发送图片识图自动回传、模型自主截图识图工具,支持多协议与一键安装/卸载。
Auto-blame: when a turn closes, asks an LLM for three cynical follow-up prompts and shows them as click-to-send bubbles above the composer. Host-gated master toggle in the settings page.
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.