dsh-memory-ga
Run the following command in DeepSeek Harness:
dsh plugin install DiligenceLai/dsh-memory-ga
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install DiligenceLai/dsh-memory-ga in the DeepSeek Harness terminal to install the plugin; the full source repository is at https://github.com/DiligenceLai/dsh-memory-ga .
About this plugin
Coding agents suffer a frustrating blind spot: they forget. Meanwhile, vector memory products swing to the opposite extreme, hoarding drafts, wrong turns, and unverified guesses into a noisy store. dsh-memory-ga takes a disciplined path: long-term writes are allowed only after a tool or the user has confirmed the fact, eliminating silent memory poisoning at the source.
The core mechanics are deliberately small. Each turn hard-injects a compact L1 index (roughly 30 lines or fewer) plus your RULES, so the model never has to chase a pointer it may never open. A session notepad (key_info) is re-injected while non-empty, giving the agent a working scratchpad. The settlement protocol, triggered by start_long_term_update, hands you a structured review flow but never edits your files on its own. Everything lives as plain UTF-8 under the local memory directory: diff it, back it up, hand-edit it. No embeddings, no cloud bank, no silent retain.
Who it is for: developers who want controlled crystallization instead of automatic recall; teams that keep Skills (how-to) and L3 facts (what-is) strictly separate; and anyone who insists on git-friendly, auditable, zero-dependency memory.
Use Cases
- Keeping critical facts and rules intact across long sessions
- Distilling hard-won lessons into auditable local memory files after a session
- Reviewing memory diffs with plain-text files across concurrent projects
Best For
- Agent users who want a transparent, auditable memory layer
- Developers who prefer local plain-text files over vector databases or cloud dependency
- Teams that keep Skills and factual memory strictly separated
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.