AI Agent Hub
Back to plugins
🧠

dsh-notes

Memory Updated 2026.09.11

Run the following command in DeepSeek Harness:

dsh plugin install suomir1995/dsh-notes

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

Run dsh plugin install suomir1995/dsh-notes in the DeepSeek Harness terminal to install this plugin; the full source is hosted at https://github.com/suomir1995/dsh-notes .

About this plugin

dsh-notes solves a very concrete problem: while you work with an AI agent inside DeepSeek Harness, your ideas, to-dos, and fragments are scattered across a dozen apps that the agent cannot read from or write to. Instead of spinning up another note-taking app, this plugin pulls everything back onto your local disk—plain Markdown files grouped by directory, no database, no cloud service, and everything is still there after a restart.

The mental model is simple: a first-level subdirectory under the root is a group, and every .md file inside a group is a note. The sidebar entry opens a full web page where you browse groups, paste-create notes, edit, rename, and delete. Six agent tools (note_root, note_list, note_rules, note_create, note_read, note_delete) let the model read convention files such as AGENTS.md via note_rules before filing a note, creating a shared human-and-agent context. The root directory can be switched in-page and is remembered instantly; writes use atomic rename to prevent concurrent corruption; names are sanitized against path traversal, keeping the security boundary tight.

Who is it for? If you already code and run agents daily in DeepSeek Harness and want a lightweight, version-controllable local note space that the AI can participate in directly, dsh-notes is that just-enough solution.

Use Cases

  • Capture ideas, to-dos, and code snippets on the fly while working with an AI agent in Harness
  • Have the agent file and retrieve notes according to team convention files such as AGENTS.md
  • Organize document fragments across multiple projects into tidy directory groups

Best For

  • Developers who code and run agents daily in DeepSeek Harness
  • Engineers who want their AI agent to read and write local notes directly
  • Users who prefer a lightweight, dependency-free, file-based workflow