</>MCP Agents Market
Plugin

memsearch

by zilliztech2.6kPythonUpdated 2026-09-07

A persistent, unified memory layer for all your AI agents (e.g. Claude Code, Codex, DSH), backed by Markdown and Milvus.

Claude CodeCodexDeepSeek HarnessOpenClawOpenCode

Memsearch is a cross-platform semantic memory layer that provides persistent, searchable conversation history for AI coding agents including Claude Code, Codex, DeepSeek Harness, OpenClaw, and OpenCode. It automatically captures agent conversations into human-readable Markdown files while maintaining a Milvus vector database index for semantic search and recall. The plugin enables agents to remember past decisions, debugging sessions, and project context across sessions and platforms, with memories flowing between different AI assistants. Developers can also use its Python API and CLI to build custom memory-augmented agents.

Key Features

Automatic conversation capture across Claude Code, Codex, DeepSeek Harness, OpenClaw, and OpenCode with zero configuration
Markdown-first architecture where .md files are the source of truth and Milvus serves as a rebuildable shadow index
Hybrid search combining BM25 sparse retrieval, dense vector embeddings, and reciprocal rank fusion reranking
Three-layer progressive retrieval: search chunks, expand to full sections, then access original transcripts
Local ONNX embedding model (bge-m3) runs on CPU with no API key required, or switch to OpenAI/Ollama/other providers
Automatic skill distillation that converts repeated workflows into reusable, installable agent skills following the Agent Skills standard
Optional background maintenance that keeps PROJECT.md and USER.md notes current across sessions
Real-time file watcher with SHA-256 content hashing to skip unchanged chunks and avoid redundant embedding calls

Use Cases

  • 01Resume debugging threads by searching how similar Redis, Docker, or deployment issues were solved in past sessions
  • 02Recover decision rationale to understand why the project chose one architecture, library, or API design over another
  • 03Trace feature evolution across sessions including files changed and tradeoffs discussed
  • 04Perform code archaeology to find when and why a module, config file, or workflow was modified
  • 05Carry unified project context across multiple AI agents so Claude Code, Codex, and DeepSeek Harness share the same memory
  • 06Build custom memory-augmented agents using the Python API and CLI for RAG pipelines or harness engineering

Related Plugins

View more

andrej-karpathy-skills

A single CLAUDE.md file to improve Claude Code behavior, derived from Andrej Karpathy's observations on LLM coding pitfalls.

Plugin216.1k

claude-mem

Persistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More

Plugin95k

Understand-Anything

Graphs that teach > graphs that impress. Turn any code into an interactive knowledge graph you can explore, search, and ask questions about. Works with Claude Code, Codex, Cursor, Copilot, Gemini CLI, and more.

Plugin84.9k

rtk

CLI proxy that reduces LLM token consumption by 60-90% on common dev commands. Single Rust binary, zero dependencies

Plugin82.1k

memsearch — FAQ

What is memsearch?+

Memsearch is a persistent semantic memory plugin for AI coding agents that automatically captures conversations into searchable Markdown files. It works across Claude Code, Codex, DeepSeek Harness, OpenClaw, and OpenCode, allowing agents to recall past decisions, debugging sessions, and project context.

How do I install memsearch for Claude Code?+

Run '/plugin marketplace add zilliztech/memsearch' followed by '/plugin install memsearch' in Claude Code, then restart. The plugin captures conversations automatically after installation with no additional setup required.

Which AI clients and agents does memsearch support?+

Memsearch provides plugins for Claude Code, Codex, DeepSeek Harness (DSH), OpenClaw, and OpenCode. Memories flow across all platforms automatically, and developers can use the Python API to integrate it into custom agents.

Does memsearch require an API key or paid service?+

No API key is required by default—memsearch uses a local ONNX embedding model (bge-m3) that runs on CPU and stores data in a local Milvus Lite file. You can optionally switch to OpenAI embeddings, Ollama, or Zilliz Cloud (which offers a free tier).

What are the prerequisites for using memsearch?+

Python 3.10 or higher is required. The default configuration works immediately with local ONNX embeddings and Milvus Lite (a single-file database). On first launch, the ONNX model (~558 MB) downloads automatically from HuggingFace Hub.

Is memsearch free and open source?+

Yes, memsearch is open source under the MIT license. The default local configuration is completely free, and Zilliz Cloud offers a free tier if you prefer a managed vector database.

How do I install memsearch?+

Open the source repository on GitHub and follow its README. memsearch is a plugin — MCP Agents Market links you directly to the official repo.

Is memsearch free?+

memsearch is an open-source project hosted on GitHub. Check the repository for its license and any usage requirements.

Related searches

Knowledge & Memory PluginsDeveloper Tools PluginsProductivity & Workflow PluginsBest MCP ServersBest SkillsBest AgentsTop rated listingsClaude Code toolsBrowse all categories