Memory and retrieval
dsh-continual-evolve
Adds workspace-partitioned SQLite memory to DSH, with indexed context injection and gated single-writer updates. Extraction and write behavior depend on the plugin's configured workflow.
Memory and retrieval
Memory and retrieval cover persistent information, search, and context organization. This independent directory groups DeepSeek Harness plugins to make related projects easier to find; inclusion does not imply official endorsement. Each published entry links to its public sources and the installation information for that entry. Read the upstream documentation for supported versions, dependencies, permissions, and operational limitations before choosing a plugin. Third-party code can change after a directory observation, and a category is only a navigation aid, not a security audit or a compatibility guarantee. Check the source date and review the actual project before installing or enabling anything in your environment.
Explore these DeepSeek Harness plugins and review each source before using dsh plugin add.
Memory and retrieval
dsh-mneme
Adds cross-session structured memory to DSH with retrieval, background consolidation, and Markdown mirrors; data stays local, while offline semantic search requires local embeddings.
Memory and retrieval
dsh-mnemon
Composes DSH runtime memory, project documents, and long-term memory spaces, retrieving context as needed; it also offers optional skill management and backtesting with configurable storage.
Memory and retrieval
dsh-auto-memory
Adds cross-session memory recall, consolidation, skill crystallization, and context management to DSH; semantic retrieval tiers have additional model-component requirements.