Introduction

DeepSeek Harness (DSH) uses a plugin-based architecture, and tools in the ecosystem usually exist as “functional modules.” During development and daily use, it is necessary to stay sensitive to industry trends. dsh-hotspots is a real-time hotspot plugin designed for the DSH Web environment, aiming to address the need to quickly view up-to-date hotspot information from the sidebar.

Plugin Overview

The plugin is maintained by developer qys0012 and follows the MIT License. It provides real-time hotspot viewing through a sidebar entry. The content is mainly divided into two parts:

  • AI Highlights: covering large models, Agents, skill plugins, workflows, and free model updates.
  • Engineering Hotspots: covering intelligent manufacturing, robotics, policies and regulations, MCU × AI, and corporate updates.

Each hotspot includes key information extracted and source links. The system automatically recognizes and labels platforms (such as Douyin, Xiaohongshu, Bilibili, Zhihu, WeChat Official Accounts, 36Kr, Jiqizhixin, Quantum Bit, IT Home, government websites, etc.).

Core Features

Sidebar Entry and Three-View Panel

The plugin inserts a sidebar entry, located below the “New Conversation” button and above “Workspace”, at the same level as the task board. Clicking it expands a hotspot panel and provides switching among three views:
* All: displays all aggregated hotspots.
* AI Highlights: displays only AI-related updates.
* Engineering Hotspots: displays only engineering-related updates.
The three views automatically deduplicate and merge results, and include item counts.

Filtering and Reading Experience

The panel supports two filtering methods:
1. Keyword local instant filtering: enter a keyword in the panel to filter the list in real time.
2. Source chips filtering: automatically extract the top 8 sources by frequency; click one to filter.
For reading experience, it supports relative time display, read marks (data persisted in localStorage, with one-click clear), shortcuts (Esc closes the panel, R refreshes the list), and mouse wheel scrolling.

Error Handling and Management

When the search service is unavailable, the panel displays the actual error reason (for example, ModSearch rate-limit messages) and provides a “Copy Diagnostic Information” button.
In “Settings → Web UI Plugins section”, a management card for this plugin is displayed, enabling unified management with other Web plugins.

Installation and Enablement

As a DSH plugin package mount, before installation ensure the host process has configured web.search integration (ctx.web).

  1. Place the plugin code in the local path /path/to/dsh-hotspots.
  2. Run the installation command:
dsh plugin --profile <profile名> add link:/path/to/dsh-hotspots

After installation, the plugin automatically injects the sidebar entry.

Data Sources and Architecture

Data Flow

The data flow is one-way. The browser requests data from the host process via same-origin loopback routes, and the host process aggregates queries through web.search. This ensures no direct third-party connections and fully reuses the search capabilities already deployed.

Underlying Search and Rate Limiting

The underlying search uses ModSearch’s Firecrawl keyless free tier, limited to 1000 queries per month.
When 429 rate limiting is triggered, the panel displays a recovery countdown. Recovery methods include:
1. Wait for the rate limit to reset automatically.
2. Configure a Free API Key in the Search engine (ModSearch) card under “Settings → Plugins” (supports Firecrawl / Tavily / Exa).

Fault Tolerance Mechanism

The plugin applies circuit-breaker handling for search failures: once a failure occurs, it stops and no longer sends all 8 concurrent queries. It also uses a no-caching-on-failure strategy, ensuring the next refresh attempts to fetch data from real sources.

Uninstallation

To remove the plugin, run the following command:

dsh plugin --profile <profile名> remove hotspots

Summary

dsh-hotspots provides a lightweight sidebar hotspot aggregation solution. Through local filtering and a one-way data flow design, it balances real-time information with usability efficiency. For developers who need to follow AI industry trends or frontier engineering trends, it is an easy-to-integrate tool.

  • Plugin directory: https://www.skillhub.cn/plugins/qys0012/dsh-hotspots
  • Source repository: https://github.com/qys0012/dsh-hotspots