AI Skill Hub 推荐使用:开源MCP工具:Rust实现的MCP服务器 是一款优质的MCP工具。AI 综合评分 7.5 分,在同类工具中表现稳健。如果你正在寻找可靠的MCP工具解决方案,这是一个值得深入了解的选择。
MCP server in Rust for AI agent persistent memory:branch-aware session handoffs,提供了一个开源的MCP服务器,支持AI代理持久性内存和分支感知会话转移
开源MCP工具:Rust实现的MCP服务器 是一款遵循 MCP(Model Context Protocol)标准协议的 AI 工具扩展。通过 MCP 协议,它可以让 Claude、Cursor 等主流 AI 客户端直接访问和操作外部工具、数据源和服务,实现 AI 能力的无缝扩展。无论是文件操作、数据库查询还是 API 调用,都可以通过自然语言在 AI 对话中直接触发,极大提升生产效率。
MCP server in Rust for AI agent persistent memory:branch-aware session handoffs,提供了一个开源的MCP服务器,支持AI代理持久性内存和分支感知会话转移
开源MCP工具:Rust实现的MCP服务器 是一款遵循 MCP(Model Context Protocol)标准协议的 AI 工具扩展。通过 MCP 协议,它可以让 Claude、Cursor 等主流 AI 客户端直接访问和操作外部工具、数据源和服务,实现 AI 能力的无缝扩展。无论是文件操作、数据库查询还是 API 调用,都可以通过自然语言在 AI 对话中直接触发,极大提升生产效率。
# 方式一:通过 Claude Code CLI 一键安装
claude skill install https://github.com/edg-l/engram-mcp
# 方式二:手动配置 claude_desktop_config.json
{
"mcpServers": {
"--mcp---rust---mcp---": {
"command": "npx",
"args": ["-y", "engram-mcp"]
}
}
}
# 配置文件位置
# macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
# Windows: %APPDATA%/Claude/claude_desktop_config.json
# 安装后在 Claude 对话中直接使用 # 示例: 用户: 请帮我用 开源MCP工具:Rust实现的MCP服务器 执行以下任务... Claude: [自动调用 开源MCP工具:Rust实现的MCP服务器 MCP 工具处理请求] # 查看可用工具列表 # 在 Claude 中输入:"列出所有可用的 MCP 工具"
// claude_desktop_config.json 配置示例
{
"mcpServers": {
"__mcp___rust___mcp___": {
"command": "npx",
"args": ["-y", "engram-mcp"],
"env": {
// "API_KEY": "your-api-key-here"
}
}
}
}
// 保存后重启 Claude Desktop 生效
<p align="center"> <img src="assets/logo.svg" alt="Engram Logo" width="180" height="180"> </p>
<p align="center"><strong>Git-aware session memory for coding agents.</strong></p>
<p align="center"> <a href="https://crates.io/crates/engram_mcp"><img src="https://img.shields.io/crates/v/engram_mcp.svg" alt="Crates.io"></a> <a href="https://github.com/edg-l/engram-mcp/actions/workflows/ci.yml"><img src="https://github.com/edg-l/engram-mcp/actions/workflows/ci.yml/badge.svg" alt="CI"></a> <a href="https://github.com/edg-l/engram-mcp/actions/workflows/release.yml"><img src="https://github.com/edg-l/engram-mcp/actions/workflows/release.yml/badge.svg" alt="Release"></a> </p>
Engram is an MCP server that gives coding agents persistent memory scoped to your git branches. Capture a session as a structured handoff, switch branches, and pick up later with the prior session's decisions, blockers, and todos surfaced automatically. Backed by local SQLite and ONNX embeddings; no cloud, no LLM calls for storage.
cargo install engram_mcp
Installs engram (MCP server) and engram-cli (command-line tool).
From source:
git clone https://github.com/edg-l/engram-mcp.git
cd engram-mcp
cargo build --release
End of session, capture state on the current branch:
```bash engram-cli handoff create
| Variable | Description | Default |
|---|---|---|
ENGRAM_DB | SQLite database path | ~/.local/share/engram/memories.db |
ENGRAM_PROJECT | Project scope identifier | Git root directory name |
ENGRAM_DECAY_INTERVAL | Decay job interval (seconds) | 3600 (1 hour) |
ENGRAM_RECLUSTER_INTERVAL | Re-clustering job interval (seconds) | 21600 (6 hours) |
ENGRAM_MAX_CANDIDATES | Max candidate embeddings to score during context retrieval | 200 |
ENGRAM_MCP_TOOL_PROFILE | Advertised MCP tool surface: full (18 tools), core (11), or minimal (3: memory_context, memory_store, handoff_resume). Dispatch stays permissive — non-advertised tools still execute with a one-time [engram] warning per process | full |
ENGRAM_HOOK_DEDUP_SKIP | Similarity threshold above which hook captures are silently dropped (clamped to [0.5, 1.0]) | 0.95 |
ENGRAM_HOOK_DAILY_CAP | Max hook-captured memories per project per UTC day; 0 = unlimited | 50 |
| Tool | Description |
|---|---|
handoff_create | Capture a session handoff with structured sections (summary, decisions, todos, blockers, mental_model, next_steps, notes) |
handoff_resume | Retrieve top sections from recent handoffs on the current branch, plus linked memories |
handoff_search | Search handoff sections by content; filter by branch or section name |
memory_store | Store a memory with embedding, auto-dedup, auto-cluster, contradiction detection |
memory_query | Semantic search with hybrid scoring, pagination, branch filtering |
memory_context | Load relevant memories for a task (hierarchical retrieval via clusters) |
memory_update | Update content, tags, importance, pinned status |
memory_delete | Remove a memory and its relationships |
memory_link | Create typed relationships between memories |
memory_graph | Traverse relationship graph from a root memory |
memory_store_batch | Store up to 100 memories atomically |
memory_delete_batch | Delete multiple memories by ID |
memory_export | Export project memories to JSON |
memory_import | Import from JSON (merge or replace modes) |
memory_stats | Project statistics (counts, types, pinned, global, clusters) |
memory_prune | Remove low-relevance memories (dry run by default) |
memory_dedup | Find and merge duplicate memories (dry run by default) |
memory_promote | Promote a branch-local memory to global scope |
| Tool | Branch-aware | Structured handoffs | Per-section embeddings | Local-only | Storage |
|---|---|---|---|---|---|
| **Engram** | ✓ | ✓ | ✓ | ✓ | SQLite + ONNX |
| [mcp-memory-service](https://github.com/doobidoo/mcp-memory-service) | — | — | — | ✓ | SQLite-vec |
| [Mem0](https://mem0.ai) | — | — | — | partial | Vector DB + LLM extraction |
| [Zep / Graphiti](https://getzep.com) | — | — | — | — | Neo4j |
| [Letta (MemGPT)](https://letta.com) | — | OS-style blocks | — | partial | Pluggable |
Engram is opinionated for coding agents using git; the others target broader assistant memory. Comparison as of 2026-05.
该项目是一个开源的MCP服务器,提供了AI代理持久性内存和分支感知会话转移的功能,代码质量良好,但缺乏文档和示例
AI Skill Hub 为第三方内容聚合平台,本页面信息基于公开数据整理,不对工具功能和质量作任何法律背书。
建议在沙箱或测试环境中充分验证后,再部署至生产环境,并做好必要的安全评估。
✅ Apache 2.0 — 宽松开源协议,可商用,需保留版权声明和 NOTICE 文件,含专利授权条款。
总体来看,开源MCP工具:Rust实现的MCP服务器 是一款质量良好的MCP工具,在同类工具中具备一定竞争力。AI Skill Hub 将持续追踪其更新动态,建议收藏备用,结合自身场景选择合适时机引入使用。
| 原始名称 | engram-mcp |
| 原始描述 | 开源MCP工具:MCP server in Rust for AI agent persistent memory: branch-aware session handoffs。⭐7 · Rust |
| Topics | mcpagent-memoryai-agentsanthropicclaudeembeddingsrust |
| GitHub | https://github.com/edg-l/engram-mcp |
| License | Apache-2.0 |
| 语言 | Rust |
收录时间:2026-06-04 · 更新时间:2026-06-08 · License:Apache-2.0 · AI Skill Hub 不对第三方内容的准确性作法律背书。
选择 Agent 类型,复制安装指令后粘贴到对应客户端