[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-94900":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":15,"subscribersCount":15,"size":15,"stars1d":15,"stars7d":16,"stars30d":16,"stars90d":15,"forks30d":15,"starsTrendScore":15,"compositeScore":17,"rankGlobal":10,"rankLanguage":10,"license":18,"archived":19,"fork":19,"defaultBranch":20,"hasWiki":21,"hasPages":19,"topics":22,"createdAt":10,"pushedAt":10,"updatedAt":26,"readmeContent":27,"aiSummary":28,"trendingCount":15,"starSnapshotCount":15,"syncStatus":29,"lastSyncTime":30,"discoverSource":31},94900,"dsh-context","bowenliang123\u002Fdsh-context","bowenliang123","A DeepSeek Harness plugin for  Context dashboard and command — showing what the model's context window is made of and how it evolves.","https:\u002F\u002Fwww.npmjs.com\u002Fpackage\u002Fdsh-context",null,"TypeScript",592,5,105,0,487,76.33,"Apache License 2.0",false,"main",true,[23,24,25],"dsh-external","dsh-plugin","dsh-plugins","2026-08-24 04:01:22","\u003Cp align=\"center\">\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fbowenliang123\u002Fdsh-context\u002Fmain\u002Fdocs\u002Fsocial-preview.png\" width=\"840\" alt=\"Social preview\">\u003C\u002Fp>\n\n# dsh-context\n\n[![npm version](https:\u002F\u002Fimg.shields.io\u002Fnpm\u002Fv\u002Fdsh-context)](https:\u002F\u002Fwww.npmjs.com\u002Fpackage\u002Fdsh-context)\n[![GitHub stars](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002Fbowenliang123\u002Fdsh-context?style=social)](https:\u002F\u002Fgithub.com\u002Fbowenliang123\u002Fdsh-context)\n\n**The best [DeepSeek Harness plugin](https:\u002F\u002Fwww.deepseek.com\u002Fharness\u002F) for Agent's context insights and management.**\n\n`dsh-context` provides full context lifecycle management features.\n- **Context tab** — an UI context dashboard for DeepSeek Harness’s context stats, composition, history, events, and messages.\n- **`\u002Fcontext` command** — the slash command shows the context model for current context composition and recent context evolution.\n\n## Install \u002F Update\n\nTo Install from any DeepSeek Harness installation:\n\n```sh\ndsh plugin --profile web add dsh-context\n```\n\nOr to update the `dsh-context` plugin:\n\n```sh\ndsh plugin --profile web update dsh-context@latest\n```\n\nThen start the web UI with `dsh web`. No build step, no restart.\n\n## Use it\n\n### Context tab\n\nOpen any session and click the **Context \u002F 上下文** tab:\n\n![Context panel overview](https:\u002F\u002Fraw.githubusercontent.com\u002Fbowenliang123\u002Fdsh-context\u002Fmain\u002Fdocs\u002Fcontext-overview.png)\n\n### ⌨️ `\u002Fcontext` command — In-session Context Insight modal\n\nType `\u002Fcontext` (or pick it from the `\u002F` menu) and press Enter: a centered dialog shows the provider-anchored occupancy headline, the six-category composition bar, and the last-10-turn trend chart — hover or click a bar for its full breakdown, exactly like the tab.\n\n![Context command](https:\u002F\u002Fraw.githubusercontent.com\u002Fbowenliang123\u002Fdsh-context\u002Fmain\u002Fdocs\u002Fcontext-command.png)\n\n## What you'll see\n\n### 📊 Context stats — the session at a glance\n\nTurns, steps, how many injections, compactions, and prunes have happened.\n\n### 🧱 Current composition — what's in the window right now\n\nA six-color stacked bar scaled against the model's full context window (the gray track is your remaining headroom): system prompt, tool schemas, your messages, injected context, assistant replies, and tool results — plus the top-5 most expensive tool schemas. When a conversation starts degrading, this is where you find out *which part ate the budget*.\n\n### 📈 History — watch the window grow (and get compacted)\n\nOne stacked bar per model request, finer than per-message. Toggle between **Turn** and **Step** granularity, scroll sideways through the session, hover any bar for a quick tooltip, and click to pin the full breakdown — including provider-reported actual prompt\u002Foutput tokens next to the estimate. **✂ marks where compaction or pruning happened** — watch the bars drop:\n\n![History chart with a pinned request](https:\u002F\u002Fraw.githubusercontent.com\u002Fbowenliang123\u002Fdsh-context\u002Fmain\u002Fdocs\u002Fhistory-detail.png)\n\nAbove: a real session that grew to ~563k tokens across 48 turns, then compaction (✂) recycled −535.5k in one step, and the conversation continued from a fresh, small window.\n\nIn **Step** granularity, hovering any bar shows that single step's context info instantly — its turn\u002Fstep, timestamp, and estimated vs. provider-reported token counts:\n\n![History chart with a step hover tooltip](https:\u002F\u002Fraw.githubusercontent.com\u002Fbowenliang123\u002Fdsh-context\u002Fmain\u002Fdocs\u002Fhistory-step-hover.png)\n\n### ⚡ Context events — when and why the window changed\n\nEvery compaction, tool-output prune, skill or plugin context injection, and model switch — each with its token delta, turn\u002Fstep attribution, and timestamp. Filter by category (**Inject \u002F Compact \u002F Prune \u002F Switch**) to see exactly when each kind of event happened and its impact — e.g. when a skill was injected, when instructions were added, or how much a compaction reclaimed:\n\n![Context events and messages](https:\u002F\u002Fraw.githubusercontent.com\u002Fbowenliang123\u002Fdsh-context\u002Fmain\u002Fdocs\u002Fcontext-events.png)\n\n### 💬 Messages — the currently model-visible surface\n\nThe exact message list the model sees right now, newest first, with a per-message token cost.\n\n## Like it?\n\nIf `dsh-context` helped you understand what your agent is carrying around, a ⭐ on [GitHub](https:\u002F\u002Fgithub.com\u002Fbowenliang123\u002Fdsh-context) is much appreciated — and issues\u002FPRs are welcome!\n\n## License\n\n[Apache-2.0](https:\u002F\u002Fwww.apache.org\u002Flicenses\u002FLICENSE-2.0)\n","dsh-context 是一个为 DeepSeek Harness（DSH）设计的上下文可视化与管理插件，用于实时监控和分析大模型推理过程中的上下文窗口构成与动态演化。它提供 Web UI 上下文仪表盘和 `\u002Fcontext` 命令弹窗，支持按类别（系统提示、工具 Schema、用户消息、助手回复等）展示当前上下文占用比例，并以时间轴形式呈现每轮请求的 token 消耗、压缩与裁剪事件。基于 TypeScript 开发，开箱即用，无需重启服务。适用于需要调试长对话性能瓶颈、优化上下文编排策略或验证工具调用开销的 AI Agent 开发与运维场景。",2,"2026-08-17 02:30:13","CREATED_QUERY"]