AI Agent Hub
Back to plugins
🧠

dsh-plugin-memory

Memory Updated 2026.09.10

Run the following command in DeepSeek Harness:

dsh plugin install menotbobbybrown/dsh-plugin-memory

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

Run dsh plugin install menotbobbybrown/dsh-plugin-memory in your terminal to install the plugin; the full source code is available at https://github.com/menotbobbybrown/dsh-plugin-memory

About this plugin

Every new DeepSeek Harness conversation starts from zero. The agent does not remember the architecture decisions you locked in last week, the coding style you prefer, or the conventions your team agreed on. dsh-plugin-memory exists to close that gap: it gives Harness agents persistent long-term memory that survives across sessions and workspace lifecycles, so context actually sticks.

Its core capabilities span three areas. First, long-term fact and preference storage lets the agent retain your coding habits, technology choices, and project norms. Second, combined semantic and keyword recall ranks results by relevance so the right context is pulled for the current prompt. Third, Cordis-native memory_reminder and memory_recall tools are exposed to the agent automatically, and a dsh-memory CLI lets you add, search, list, and forget entries straight from the terminal.

This plugin is built for developers who work with DeepSeek Harness on multi-project or multi-week tasks and want their agent to carry context forward without re-explaining everything every session. If you are tired of repeating your preferences and project background, this is worth installing.

Use Cases

  • Agent retains coding preferences and architecture decisions across sessions without re-explaining
  • Quickly search, add, or forget memory entries from the terminal via dsh-memory CLI
  • Keep each project's workspace conventions and context alive while juggling multiple workspaces

Best For

  • Developers building multi-week projects with DeepSeek Harness
  • Tech leads who want agents to respect team coding conventions and personal preferences
  • Full-stack engineers maintaining a single codebase over months and needing continuous context