AI Agent Hub
Back to plugins
🖥️

dsh-plugin-todo-scanner

Client Updated 2026.09.10

Run the following command in DeepSeek Harness:

dsh plugin install zhaoxuejie/dsh-plugin-todo-scanner

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

Install this plugin in DeepSeek Harness by running dsh plugin install zhaoxuejie/dsh-plugin-todo-scanner; the source is available at https://github.com/zhaoxuejie/dsh-plugin-todo-scanner

About this plugin

TODO, FIXME, HACK, and OPTIMIZE comments pile up in every codebase, scattered across hundreds of files and rarely revisited in any structured way. dsh-plugin-todo-scanner turns that ambient comment noise into a single actionable checklist: it recursively scans a local project directory for eight common marker types, extracts each hit with file, line number, and two lines of surrounding context, and rolls everything into per-type, per-file, and per-language statistics that an LLM can query and reason over directly.

On the host side the plugin registers eight tools covering the full lifecycle—scan, list, detail, single update, batch update, Markdown export, statistics, and clear—so the model can filter, sort, read context, transition status (pending / in_progress / done / ignored), and clear results within a single conversation. On the client side a dedicated TODO Radar panel card surfaces stat cards, a type-distribution chart, a filterable and sortable list, and inline status actions in one view. One-click Markdown export grouped by type or by file makes it easy to drop the checklist into a weekly report, a GitHub issue, or a team board.

The scanner is strictly read-only, protected by a whitelist plus a system-directory blacklist with hard caps on file count and size. All state lives in memory, is isolated per session, and is wiped on unload—nothing is written back into the project. It is well suited to individual developers and small teams already working inside the DeepSeek Harness (Cordis) ecosystem, and to AI-assisted workflows where an LLM should automatically surface legacy TODOs during code review or refactoring.

Use Cases

  • Surface outstanding TODO, FIXME, and HACK markers during code review
  • Triage all pending code markers before a refactoring sprint
  • One-click export of unfinished tasks into a Markdown checklist for reports or Issues

Best For

  • Developers working in the DeepSeek Harness (Cordis) ecosystem
  • Small teams leveraging AI-assisted code review and refactoring
  • Solo developers who want ongoing visibility into legacy code markers