DeepSeek Harness Plugins · Memory
Plugins that extend DeepSeek Harness (DSH) with vision, workflows, memory, web tools and more.
A literature-terms memory plugin that lists term entries and retrieves details by id or name, built in pure Node with no network dependency.
A pure Node.js memory plugin with zero network dependencies, built for lightweight and clean operation.
Tracks which files the Harness agent edited, with which model and session, and uses git-ai to attribute code line-by-line in your editor.
Intercepts delete commands (rm, Remove-Item, etc.) and turns every deletion into a recoverable trash move, with one-click restore and a single confirmed purge path.
A DeepSeek Harness plugin that previews old conversations by age threshold and archives them in one click, non-destructively with logs retained and restorable.
Migrate an existing DeepSeek Harness session to another registered workspace while preserving its session ID and conversation history.
Regex search over DSH conversation (assistant text, reasoning, user messages, tool results) with match statistics, a paginated highlighted table, and a model-invokable pattern_search tool.
A lightweight correction-delivery plugin for DeepSeek Harness: record, explicitly adopt, and recall corrections within scoped profiles, with log reconstruction and no external dependencies.
Deterministically injects two-level memory (global and project MEMORY.md plus daily notes) into every new session context with an independent byte budget, removing reliance on the model to self-read.
A file-backed Markdown notes workspace in the DSH sidebar with live editing, GFM rendering, autosave, and search — keep chat insights as persistent local .md files.
Cleans up accumulated sub-agent sessions in DSH via self-cleanup and global sweep tools, with reversible archiving or permanent deletion.
Bridge the local codebase-memory-mcp knowledge-graph server into DSH, exposing 14 tools for code search, path tracing, and change detection with hot-reloadable config and live status monitoring.