dsh-cc-haha-dream
Run the following command in DeepSeek Harness:
dsh plugin install yihefeikong-rgb/dsh-cc-haha-dream
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install yihefeikong-rgb/dsh-cc-haha-dream in your terminal; the source is at https://github.com/yihefeikong-rgb/dsh-cc-haha-dream
About this plugin
dsh-memory captures useful snippets as you chat, but over time those notes accumulate duplicates, stale entries, and contradictions, while the MEMORY.md index grows past any manageable size. dsh-dream handles the bedtime tidy-up: a background process that periodically reviews recent sessions and existing memories, merges duplicates, refreshes outdated entries, removes contradictions, and trims the MEMORY.md index back to 200 lines or fewer. The consolidation runs silently in the background without blocking your conversation, and the result appears as a single system message.
Under the hood, the plugin forks a sub-agent that works through a four-phase prompt (orient, gather recent signals, consolidate, trim the index) and emits a strict JSON action list of create, merge, update, and delete operations. On the host side, every action is validated in a batch before any write begins; all file operations are restricted to the memory directory; deletions and overwrites are preceded by automatic .history/ snapshots; and a .dream-lock file prevents concurrent runs. Triggering is flexible: by default it fires once every 24 hours after at least five new sessions have accumulated, but you can also run it on demand with the /dream command or the memory_dream tool.
Who it is for: DSH users who already rely on dsh-memory for everyday capture. dsh-dream does not modify dsh-memory at all; the two plugins share the same ~/.dsh/memory/ directory and are fully compatible with the existing file format and MEMORY.md index. Through the settings panel or a patch file you can tune the trigger mode (time, content volume, or hybrid), scope, digest caps, sub-agent model, provider, and timeout so the consolidation cadence fits your workflow.
Use Cases
- Session notes have grown large with duplicates, stale entries, and contradictions
- MEMORY.md index exceeds 200 lines and search efficiency drops
- Want background memory tidying without interrupting the current conversation
- Periodically distill new information from recent sessions and merge it into existing memories
Best For
- DSH users already relying on dsh-memory for daily capture
- Developers who need a lean, easy-to-search memory store
- Users who want automated index maintenance without manual tidy-up
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.