dsh-memory-system
Run the following command in DeepSeek Harness:
dsh plugin install zhujunpeng12/dsh-memory-system
Paste the following prompt into your AI chat to install this plugin:
To install this plugin in DeepSeek Harness, run the command: dsh plugin install zhujunpeng12/dsh-memory-system, or visit the open-source address at https://github.com/zhujunpeng12/dsh-memory-system for more details.
About this plugin
DeepSeek Harness sessions are forgetful by default, forcing Agents to rebuild context repeatedly and hurting efficiency. dsh-memory-system solves this with a six-layer memory loop, engineering memory management: automatic hot-memory injection (≤14KB) per new session, on-demand cold recall (≤4.2KB) with Chinese BM25 search, and local Markdown storage for transparency and auditability. Writes use lease locks and transactions with default preview confirmation, preventing unauthorized changes, while governance and trajectory review enable an evolution loop.
Its core strength lies in local-first, zero-dependency design: no databases or vector services needed, using Python's standard library and Markdown for full privacy control. Chinese recall uses bigram BM25 and metadata reranking for precise matching; write safety employs single-writer locks and before-image protection, ideal for auditable scenarios. It's not a black-box vector store but an explainable, controllable memory system.
This plugin suits individuals or small teams in DSH workflows who prioritize local data, Chinese handling, and write security. If you seek auditable memory management, want Agents to maintain coherent cross-session context, and prefer avoiding external services, it's an excellent fit. It's not designed for multi-tenant high-frequency writes or fully automated approval-free scenarios, emphasizing user authorization and human governance.
Screenshots
Use Cases
- Need an Agent to retain project progress, user preferences, and key decisions across sessions.
- Working with codebases containing Chinese documentation or proper nouns that require accurate local history recall.
- Adding persistent memory to a DeepSeek Harness workflow in an environment that prioritizes data privacy and auditability.
Best For
- Independent developers or small teams seeking to enhance Agent context continuity.
- Individual users who prioritize data sovereignty and manage knowledge with Markdown or Obsidian.
- DSH workflow users who need to handle Chinese content and seek reliable, explainable recall effects.
Related Plugins
Traceable, searchable cross-session memory for AI agents that turns conversation knowledge into a typed knowledge graph and recalls relevant subgraphs instead of replaying full history, natively integrated with DeepSeek Harness.
Gives DSH AI cross-session long-term memory, to-do and skill management, plus multi-session orchestration, external AI delegation, and an infinite canvas that grows with you.
dsh-mnemon is a three-tier, pluggable, Agent-driven memory system for DeepSeek Harness, combining Runtime memory, Project Documents, and replaceable Memory Spaces with nine long-term providers.
An opinionated, zero-infrastructure file-based memory protocol: human-readable markdown files plus a bounded always-loaded index, curation discipline, and no database or embeddings.