@deepseek-ai/dsh-plugin-memory

Configuration

Deploy via a DSH plugin entry (see package.json exports):

- id: memory
  name: '@deepseek-ai/dsh-plugin-memory'
  config:
    enableEntryInjection: true     # prepend the how-to-save + index block each session
    enableRelevance: true          # append relevant topic files per step (data-role=memory)
    enableExtraction: true         # idle-time LLM auto-extraction
    maxRelevant: 5                 # max files surfaced per step
    relevanceTopK: 8               # max candidates the LLM selector may pick from
    relevanceBudgetChars: 2000     # per-topic char cap fed to relevance/selector
    extractionDebounceMs: 60000    # idle debounce before a pass runs
    extractionLookback: 40         # recent events scanned for a pass
    llm:                           # optional; omit provider/model to use keyword-only relevance + no extraction
      provider: deepseek
      model: deepseek-chat
      maxTokens: 1024

Without an llm route, the plugin still provides index+topics, entry injection, keyword relevance, the agent tools, and profile rotation — only LLM-based extraction and LLM relevance ranking are disabled.