dsh-auto-memory
Run the following command in DeepSeek Harness:
dsh plugin install 1304836815/dsh-auto-memory
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install 1304836815/dsh-auto-memory in your terminal to install this session-memory plugin, then restart DSH and configure it under Settings > Plugins > AutoMemory; full source code at https://github.com/1304836815/dsh-auto-memory
About this plugin
The most annoying part of a long, multi-session build is not writing code—it is losing the thread. Close the terminal, reopen the project, and the architecture decisions, constraints, and trade-offs from yesterday are gone. DSH treats every session like a fresh page, leaving you to carry context entirely in your head. dsh-auto-memory plugs exactly that gap: at the start of each session it loads the project-root memory index and hands the key conclusions straight to the model, so the conversation never truly starts from scratch.
The plugin is deliberately model-free and keeps its scope tight. While a session is running, it checks the line count and byte size of MEMORY.md against the soft limits you configure in the GUI; when you exceed either, it nudges you to tidy up your notes. If a previous session was closed without finishing the wrap-up, it drops a flag file under .dsh/ and reminds you to complete it on the next launch, then clears the flag once you are done. On first run it scaffolds the MEMORY.md index and a memory/_TEMPLATE.md note out of the box, so there is no manual bootstrapping. Every setting—index file path, line ceiling, size ceiling, enable toggle—lives in the panel under Settings > Plugins > AutoMemory, with sliders and dropdowns instead of YAML.
If you use DSH for projects that span days and multiple topics, and you prefer design decisions recorded in the repo rather than pasted by hand every time, this plugin adds that lightweight auto-remember layer with zero model dependencies. Install it, restart DSH, and the memory loop just starts working.
Use Cases
- Long multi-session projects where context is lost between sessions and the memory index needs automatic reload
- MEMORY.md index exceeds the configured line or size ceiling and the plugin proactively prompts cleanup
- A previous session was terminated without completing the wrap-up, and the next launch reminds you to finish it
Best For
- Engineers running long, multi-day projects in DSH
- Developers who prefer recording design decisions in the repo rather than relying solely on memory
- Users who prefer GUI panel configuration over editing YAML files manually
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.