Bundle
dsh-conversation-landmarks
Fixed rail of conversation landmarks with hover previews and click-to-jump navigation for DeepSeek Harness Web
- Source
- mantonlove
- stars
- 5 stars
- License
- MIT
- Updated
- Updated 3 days ago
Readme
# Conversation Landmarks for DeepSeek Harness English | [简体中文](README.zh.md) Conversation Landmarks is an independent [DeepSeek Harness](https://github.com/deepseek-ai/deepseek-harness) Web plugin for navigating long conversations. It adds a fixed, vertically centered landmark rail to the left edge of the conversation workspace. Each line represents one direct user task rather than every individual message.  ## What it does - Keeps the complete landmark rail fixed and centered in the whole conversation column while it scrolls. - Hides the rail until at least three user tasks exist (it appears from the third line). - Selects the line nearest the pointer. The selected line becomes longest and pure white; the two lines on each side taper down (28/20/16px) to the resting length. - Shows a compact preview of the user request and the latest following Assistant answer. - Loads older history when necessary, then scrolls to the selected user message with a highlight so the jump target stays fully visible. - Uses the official session projection and `conversation.input.dock` Client slot extension points. It does not patch DeepSeek Harness source files. - Performs no plugin-owned network requests and collects no telemetry.  ## Install Requirements: DeepSeek Harness `0.1.1-rc.2` or later and its supported Node.js version. Install the prebuilt plugin into the Web profile, then restart Web: ```sh dsh plugin --profile web add "github:mantonlove/dsh-conversation-landmarks#v0.1.1" dsh web ``` The repository commits the required `lib/` artifacts, so this GitHub install does not ask pnpm to run a package build script. Open the actual URL printed after `dsh web:` in the startup output. The default is `http://127.0.0.1:3080`, but the address can differ when the port or host is configured. If you run DeepSeek Harness from its source checkout, use the same commands with `pnpm dsh`: ```sh pnpm dsh plugin --profile web add "github:mantonlove/dsh-conversation-landmarks#v0.1.1" pnpm dsh web ``` Remove the plugin with: ```sh dsh plugin --profile web remove dsh-conversation-landmarks ``` ## Prompt for an AI coding agent You can paste this into an agent that has terminal access to your DeepSeek Harness installation: > Install `github:mantonlove/dsh-conversation-landmarks#v0.1.1` into the DeepSeek Harness `web` profile with `dsh plugin`, restart `dsh web`, and tell me the actual URL printed after `dsh web:`. Do not edit DeepSeek Harness source files. ## Develop ```sh pnpm install pnpm run check ``` The package is a Cordis Service plugin. Its Host half registers the `conversationLandmarks` session projection through `ctx.sessionProjections`. Its browser half mounts through `conversation.input.dock`, then renders the fixed rail in a portal. ## Feedback Report bugs or request features in [Issues](https://github.com/mantonlove/dsh-conversation-landmarks/issues). Please search existing issues before opening a new one. ## Status This is a community project, not an official DeepSeek release. The current compatibility target is DeepSeek Harness `0.1.1-rc.2`. ## License [MIT](LICENSE)
Install
dsh plugin --profile web add github:mantonlove/dsh-conversation-landmarks
Profile: web
With the hub plugin installed, ask your agent to install it by name — it resolves the same plan shown here.
dsh plugin --profile web add github:stvlynn/dsh.fish#path:packages/dsh-plugin-hub
install dsh-conversation-landmarks from the hub
- This package builds from source on install. pnpm will ask you to allow its build script — that is permission to run the package’s code on your machine, outside the agent sandbox. Only allow sources you trust.
- This source has no pinned commit, so a later push upstream changes what installs. Prefer pinning a commit.