dsh-memory
Run the following command in DeepSeek Harness:
dsh plugin install mnbvcxzaqwertyuioplm/dsh-memory
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install mnbvcxzaqwertyuioplm/dsh-memory in the DeepSeek Harness terminal to install the plugin; the full source repository is https://github.com/mnbvcxzaqwertyuioplm/dsh-memory — restart DSH web after installation to activate the bundle layer.
About this plugin
Your DSH agent forgets everything between sessions—preferences you mentioned last week, agreements you reached, project context—gone the moment you close the page. dsh-memory adds a durable, semantically searchable memory layer using a few hundred bytes of SQLite and one embedding call per write, with no extra vector database or sidecar containers to babysit.
Two tools do the heavy lifting: memory_add auto-vectorizes on write and runs a cosine dedup pass (entries with similarity >= 0.9 merge into the existing row, keeping the newest wording); memory_search ranks by embedding cosine similarity first, then blends in keyword-hit weighting, falling back to pure bigram matching when embeddings are unavailable, with an 8-second timeout so a bad network never hangs the session. Type /mem followed by a question in the composer and the agent retrieves memory first, renders results as a collapsible card, then answers from that context.
Built for developers and personal users running DSH web or headless profiles who want their agent to genuinely remember without spinning up another service. Every config option is optional, a missing API key or database degrades gracefully, and the plugin is designed so it can never crash the DSH tree on startup.
Use Cases
- Agent retains user preferences, project context, and agreements across sessions instead of starting from scratch
- Semantic recall with SQLite and one embedding call per write, no vector database or sidecar containers required
- Type /mem plus a question in the composer to force a memory search before the agent answers, ensuring no prior context is missed
Best For
- Developers running DSH web or headless profiles who want their agent to genuinely remember across sessions
- Personal users who prefer zero extra infrastructure over deploying a vector database or sidecar containers
- DSH custom-config setups that require a lightweight persistent memory layer with strict startup stability guarantees
Related Plugins
Memory layer for coding agents that indexes local session history and auto-recalls relevant context before edits, commands, and failures, with no manual search needed.
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.