dsh-memory
Run the following command in DeepSeek Harness:
dsh plugin install FuRongJun-1999/dsh-memory
Paste the following prompt into your AI chat to install this plugin:
Install in DeepSeek Harness with `dsh plugin install FuRongJun-1999/dsh-memory`; full source code: https://github.com/FuRongJun-1999/dsh-memory
About this plugin
Most agents today wake up without a continuous self: they cannot carry an identity across sessions, cannot learn from experience after weights are frozen, and cannot show auditable behavioral boundaries. dsh-memory approaches this as an AGI-grade long-term memory substrate rather than a simple key-value memory plugin. It brings LingShu (AEIS)'s spatiotemporal memory graph, five-layer memory (anchor/structural/knowledge/contextual/self), and a knowledge flywheel into DeepSeek Harness, letting agents recall/search/timeline across conversations and continuously refine experience through a verify → induce → associate → distill → reason loop.
A defining feature is white-box intelligence. Knowledge queries go through a deterministic pipeline of condition routing, compositional generation, self-verification, and knowledge consolidation — 100% LLM-free by default, with LLM kept only as an external validator. This makes answers explainable, traceable, and auditable. Together with the guardrail charter, full event logging, metacognitive tools (cognition, self_reliability, emotional_bias, recursive_reflect), and the v3.3 roleplay engine, it is suited to long-horizon role consistency and verifiable behavior constraints.
The audience is deliberately narrow: AGI researchers and people working on interpretability, protocol engineering, memory mechanisms, or role theory — not everyday end users. If you need a runnable, auditable, reproducible substrate for long-term memory and white-box intelligence experiments, dsh-memory is that infrastructure.
Screenshots
Use Cases
- Provide cross-session memory and self-continuity for agents
- Answer knowledge queries through a zero-LLM white-box pipeline
- Maintain consistent roleplay personas over long conversations
Best For
- AGI and interpretability researchers
- Protocol engineering and memory mechanism researchers
- Agent developers needing auditable long-term memory
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.