DeepSeek Harness Plugins · Memory
Plugins that extend DeepSeek Harness (DSH) with vision, workflows, memory, web tools and more.
A local memory plugin providing storyboard script templates and terminology, built in pure Node.js with no network or external service dependencies.
Monitors the system clipboard, retains a searchable history of up to 500 entries, provides a web panel for management, and exposes clipboard tools to the Agent.
A local, zero-network memory reference plugin based on Gestalt principles of visual perception, implemented purely in Node.
Git-based multi-device bidirectional sync for DSH workspace conversations and files, with conversation-only and full-file modes, path-agnostic mirroring, and one-click cloning for new devices.
A pure Node.js photography reference plugin providing photographic knowledge memory with no network or external service dependencies.
List and manage DSH historical sessions in Settings, with type-to-confirm physical deletion and audit logging.
A pure Node grid and spacing system reference plugin with no network or external service dependencies, providing layout tools for sessions.
A memory plugin referencing composition rules, built purely in Node.js with no network or external service dependencies.
Adds a shared PROGRESS.md to every DSH workspace so new sessions auto-read project state and completed sessions auto-append progress, fully local with no network calls.
Local-first context badge for DSH with per-session notes, todos, and an optional macOS foreground-window companion; no network or model calls.
Observes session metrics and injects deterministic handoff suggestions when thresholds are crossed, then delegates execution to the harness-automation CLI after approval.
Select any part of an AI answer to ask a follow-up in an isolated, persistent child session without interrupting the main conversation.