AI Agent Hub
Back to plugins
dsh-plugin-jinji preview

dsh-plugin-jinji

Memory Updated 2026.08.20

Run the following command in DeepSeek Harness:

dsh plugin install quan2005/dsh-plugin-jinji

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

Run dsh plugin install quan2005/dsh-plugin-jinji in the DeepSeek Harness terminal; the plugin source is available at https://github.com/quan2005/dsh-plugin-jinji.

About this plugin

DeepSeek Harness excels at reasoning and drafting, yet once a session ends the context is gone. The Jinji Memory Panel embeds a dual-track memory system directly into the DSH sidebar with zero third-party dependencies, no database, and no build step. Open the panel and your journal logs and entity profiles are right there, stored as plain Markdown files.

Dual-track memory works on two levels. Flow memory turns voice notes, screenshots, and pasted text into AI-structured daily event streams, browsable as a monthly card waterfall with instant multi-keyword search and full keyboard navigation. Profile memory continuously distills independent entity archives person identity, decision patterns, product positioning from those events, grouped by Self, Product, and Team, and feeds them back into future conversations. Even hundreds of entries stay instant thanks to fingerprint-cached indexing and progressive rendering.

Two features make the LLM genuinely remember: every new session auto-injects recent journal summaries and all profiles into context, and a one-click Journal-Secretary agent preset lets any session proactively write memories following a strict convention. The entire system is pure text with the LLM as its core organizer, built for anyone who wants an ever-growing, zero-toolchain memory inside DSH.

Screenshots

Use Cases

  • Browse daily journals and entity profiles directly from the DSH sidebar
  • Auto-inject memory summaries at session start so the AI recalls past discussions
  • Capture fragmented ideas via voice or text and let the AI structure them into logs
  • Track person decision patterns and product iteration context that feed back into future conversations

Best For

  • Developers who need persistent memory inside DeepSeek Harness
  • Users who manage knowledge with plain text and want to avoid databases
  • People who want the AI to proactively build person and product profiles