memorylake-ai/memorylake-harness#dsh-plugin
Memory Lake as a persistent memory layer for dsh: memory_search, memory_remember, and memory_forget tools over the memorylake CLI, a session status line, and guided setup/diagnostic skills, sharing one ~/.memorylake identity and memory set with the Claude Code and Codex plugins.
Connects DeepSeek Harness to Memory Lake, a cross-device long-term memory service. The dsh plugin supplies memory tools, prompt guidance and session status, and reads/writes the same memory tree as the Claude Code, Codex and opencode harnesses in the same repo — recall on read, sync on write. Because the identity and data tree are shared, a fact learned in one agent client is available from the others after a single setup.
Install
dsh plugin --profile web add @memorylake/dsh-pluginREADME DeepSeek Harness section: install the plugin, start dsh web, then type /memorylake-init in a session (machines already set up for the Claude Code or Codex harness can skip that last line). Part of memorylake-harness, which ships one client per agent while sharing a single identity and data tree at ~/.memorylake/harness/.
Compatibility
DSH web profile. Two Memory Lake deployments with separate accounts: memorylake.ai (international) and memorylake.cn (China); the CLI defaults to the international endpoint. All harnesses share ~/.memorylake/harness/, so one configuration serves every client.
Details
- Repo: memorylake-ai/memorylake-harness#dsh-plugin
- Category: Memory
- Stars: 0
- Version: 0.1.0 (npm @memorylake/dsh-plugin; registry repository field → github.com/memorylake-ai/memorylake-harness, verified 2026-09-11)
- Last push: 2026-09-17
- First seen: 2026-07-22
Recent updates
README tracks per-harness status: claude-plugin, codex-plugin, dsh-plugin and opencode-plugin are all marked working, with dsh-plugin published to npm as @memorylake/dsh-plugin.
FAQ
- How do I install the Memory Lake plugin for dsh?
- Run: dsh plugin --profile web add @memorylake/dsh-plugin, start dsh web, then type /memorylake-init in a session.
- Do I need to configure Memory Lake again for each agent?
- No — the harnesses share one identity and data tree at ~/.memorylake/harness/, so configuring once covers Claude Code, Codex, dsh and opencode; init can also be skipped if another harness is already set up.
- Which Memory Lake endpoint does the dsh plugin use?
- There are two deployments with separate accounts — memorylake.ai (international) and memorylake.cn (China). The CLI defaults to the international endpoint.