AI Agent Hub
Back to plugins
🖥️

dsh-typing-insight

Client Updated 2026.08.28

Run the following command in DeepSeek Harness:

dsh plugin install SYeShare/dsh-typing-insight

Paste the following prompt into your AI chat to install this plugin:

In a DeepSeek Harness chat box, type dsh plugin install SYeShare/dsh-typing-insight to install; the project source is at https://github.com/SYeShare/dsh-typing-insight

About this plugin

How many words did you actually type today? Which apps consumed your focus? What topics kept surfacing across your conversations? dsh-typing-insight turns those fuzzy questions into verifiable daily and weekly numbers on DeepSeek Harness Desktop (Windows x64). A global keyboard hook captures keystrokes while a UI Automation bridge reads committed text from the focused control; everything lands in a local SQLite database with zero runtime network calls.\n\nBeyond raw counts, the plugin computes active time, typing speed, and per-app and per-hour distributions, then layers on keyword extraction, topic clustering (up to five clusters), and a local extractive summary so you can skim a week of context in seconds. Daily reports export to Markdown, JSON, or CSV with a SHA-256 fingerprint attached. App-usage tracking reports foreground, foreground-active, and strong-interaction minutes side by side; the AFK threshold is configurable, and unknown gaps are never force-attributed to a single app. Privacy is a hard constraint: password fields block both keys and text, unknown controls are treated as passwords, and PII such as phone numbers, ID numbers, bank cards, and email addresses is masked before it ever touches disk. The plugin ships in a paused state until you explicitly review the capture scope and authorize it from the settings panel.\n\nIt is built for developers and knowledge workers who juggle Chrome, Edge, DingTalk, DSH Desktop, and other day-to-day tools and want a quantitative picture of their focus and output without surrendering keystroke-level data to any third party. Note that it targets Windows x64 only, and WeChat 4.1.x (MMUI closed renderer) records keys only with no text capture, a limitation surfaced as a badge in the panel.

Use Cases

  • Check daily word count, active time, and typing speed to quantify personal input output
  • Generate a weekly report and export to Markdown/JSON/CSV to review top topics and frequent apps
  • Compare foreground and strong-interaction minutes across apps to see where attention truly goes

Best For

  • Developers and knowledge workers switching daily among Chrome, DingTalk, DSH Desktop, and other tools
  • Freelancers who prefer data-backed reflection over gut feel to review how their time is spent
  • Privacy-sensitive users who insist on local-only storage and zero cloud upload of keystroke-level data