agentscope-ai/ReMe

AgentScope's memory kit for agents: a local-first, self-evolving personal knowledge base exposed to DSH as memory guidance, reme_search, turn capture and Auto Dream.

ReMe (Remember Me, Refine Me) is AgentScope's memory management kit: a local-first, self-evolving personal knowledge base rather than a vector store you query by hand. Inside DSH the plugin contributes configurable memory guidance, a reme_search tool, automatic capture of turns, a scheduled Auto Dream job and a ReMe Status panel, so recall is injected as context instead of being something you remember to ask for.

Memory ★ 3.7k
View on GitHub ↗

Install

dsh plugin --profile web add @agentscope-ai/reme-dsh-plugin

Install line is the one in the README's DeepSeek Harness integration table; npm package @agentscope-ai/reme-dsh-plugin, latest 0.1.2, and its registry repository field resolves back to this repo (github.com/agentscope-ai/ReMe). The plugin is the client half — it talks to a ReMe service, and the README states ReMe itself requires Python 3.11+.

Compatibility

Installs into the web profile. The README's DSH section lists the plugin as the harness-side integration for long-term-memory guidance, reme_search and turn capture; the ReMe service it calls is a separate process that needs Python 3.11 or newer.

Details

Alternatives

sens-io/memobranch · kanhaoning/NeatMem · Relistencode/dsh-recall

More plugins in Memory

Browse more in Memory

Guides for Memory plugins