bowenliang123/dsh-context
Context insight panel: see what the model's context window is made of and how it evolves — composition vs. window size, per-request history, compression/injection events, and per-message token stats.
Install
via npm
Resolves a published tarball rather than cloning the repository, and installs without any extra setup. Swap `web` for your profile name if you run another one.
via GitHub · npm package
Installing from GitHub runs the project's build script, which pnpm blocks until you allowlist it — run the command once and pnpm prints the exact key to add under `allowBuilds` in ~/.dsh/profiles/web/pnpm-workspace.yaml.
What happened when we ran it
Installed cleanly when we ran it
Every command here is run in a throwaway container against a clean profile, and the result is whatever the harness recorded — not a guess from the source. Last run yesterday.
Due diligence
What it is — dsh-context is a DeepSeek Harness plugin for context insight.
Who it is for — If you use DeepSeek Harness for multi-turn conversations, you should install it. If you use a small context window model, you can skip it.
Watch out — Sandbox testing confirmed successful registration in a new profile. No obvious issues found.
The verdict — I would install it because it shows context evolution details that matter in conversation management.
Generated by grok-4.6, and a starting point rather than a verdict. Where it says a plugin installs or does not, that is from a real run in a clean profile — everything else is read off the repository. Trust the source over this.
dsh-context is a DeepSeek Harness plugin that exposes how an agent’s context window is assembled and changes over time. It adds a Context / 上下文 tab to the web UI and a /context command that opens the same view for the current session.
The plugin presents the context model recorded for each request: occupancy against the model’s full window, a six-part composition breakdown, per-request history, compaction and pruning markers, injection and model-switch events, and per-message token costs. The history can be viewed by turn or step. Selecting a request opens its assembled context, including system prompts, tool schemas, messages, injected context, assistant replies and tool results. The panel also shows provider-reported prompt and output tokens alongside estimates where available.
dsh-context is intended for developers diagnosing context growth, compaction, injected instructions, or unexpectedly expensive tools. It depends on the DeepSeek Harness web UI and an open session, so it is not useful in a terminal-only workflow or when context inspection is unnecessary. Earlier requests may be reconstructed from the removed-message archive and marked as approximate after compaction. The README names no configuration keys, files, platform requirements, or separate data-writing behaviour.
在 DeepSeek Harness 中打开任意会话，然后选择 Context / 上下文 标签页。启动 web UI 使用 dsh web；README 说明安装或更新后无需构建步骤，也无需重启。页面展示当前会话的回合数、步骤数，以及注入、压缩和修剪的次数。
在会话中输入 /context，或从 / 菜单选择它并按 Enter。命令会打开上下文洞察窗口，其中包含当前占用量、六类组成，以及最近 10 个回合的趋势图。悬停或点击柱状图可查看详细拆分。
当前组成按模型完整上下文窗口缩放，类别包括：
历史图表按模型请求记录柱状图，可切换 Turn 和 Step 粒度，并查看估算 token 与 provider 报告的实际 prompt/output token。事件列表记录压缩、工具输出修剪、skill 或 plugin 上下文注入及模型切换，并提供 token 变化量、回合或步骤归属和时间戳。消息视图按最新优先列出当前模型可见的消息及其 token 成本。
上下文浏览器可选择 Live (next request) 或保留的步骤，展开各类别中的元素及其实际内容，包括工具描述、JSON schema、消息、reasoning、tool-call 参数和工具输出。历史图表的悬停选择会同步浏览器；展开较旧元素时，页面会自动加载更早的聊天历史。压缩前的步骤可能依据 removed-message archive 重建，并标明内容仅为近似。README 未说明额外配置、环境变量、平台范围或版本要求。
Written from the project's own documentation and kept in sync with it. Where the two disagree, the source is authoritative — read the README on GitHub
Same category
volcengine
OpenViking memory and context bundle for DeepSeek Harness: pre-step auto-recall and profile injection, session capture, `viking://` URI guarding, and recall/write memory tools backed by an OpenViking server.
No one-line install — this plugin lives inside a larger repository and publishes no npm package.
vectorize-io
Hindsight, agent memory that learns: long-term project memory with auto recall and retain, knowledge pages, deep reflection, and per-repo memory banks.
No one-line install — this plugin lives inside a larger repository and publishes no npm package.
csyangwen
为 DeepSeek Harness 带来「跨会话长期记忆 + 后台自我进化」能力的纯插件实现：五轨记忆 · git 分支感知 · 回合内自我审查 · 技能自我进化与技能管理器 · 四轨待办 · COI 调度 · 会话广播 · 会话搜索 · 提示词管理器 · 临时信息便签——零核心修改、零运行时依赖，随装随用、卸载即净。
Installed cleanly when we ran it
tinqiao-oss
The Engramory curated-memory discipline as an installable plugin ([npm: dsh-engramory](https://www.npmjs.com/package/dsh-engramory)): a deterministic 200-line / 25 KB cap on the `MEMORY.md` index via `ctx.tools.guard()` — growth denied, a shrinking rewrite always passes — plus the protocol registered as a runtime skill. The store is plain markdown, one file per fact, shared with Claude Code, Codex, Kiro, and OpenClaw.
Installed cleanly when we ran it
ZSeven-W
Noema long-term memory plugin for DSH: durable, inspectable agent memory with recall tools and a settings page.
Installed cleanly when we ran it
omdsh-dev
Cross-agent, local-first persistent memory plugin for DeepSeek Harness (DSH), powered by Mnemon. It shares long-term memory across Mnemon-enabled agents and adds runtime memory, searchable project documents, semantic recall, knowledge graph, and a Sidebar UI.
Installed cleanly when we ran it
