AI Skill Hub 推荐使用:程序内存 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。
程序内存 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。
程序内存 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/SerroAI/program-memory
# 方式二:手动配置 claude_desktop_config.json
{
"mcpServers": {
"----": {
"command": "npx",
"args": ["-y", "program-memory"]
}
}
}
# 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 程序内存 执行以下任务... Claude: [自动调用 程序内存 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": {
"____": {
"command": "npx",
"args": ["-y", "program-memory"],
"env": {
// "API_KEY": "your-api-key-here"
}
}
}
}
// 保存后重启 Claude Desktop 生效
A starter kit for building live shared program memory using Claude Code, MCP integrations, and git — no proprietary infrastructure required.
This repo covers Levels 1–3 (pull, map, loop). The proactive layer, widget layer, and graph index are out of scope. See the scope table below.
This repo contains step-by-step instructions, not code. There is nothing tonpm installordocker run. Each family folder is a guide for how to build the implementation yourself using Claude Code and native MCP integrations.
---
┌─────────────────────────────────────────────────────┐
│ Widget layer Prompt-based live views of │
│ program state (requires layer 2) │ ← out of scope
├─────────────────────────────────────────────────────┤
│ Proactive layer Autonomous loop agent monitors │
│ programs, flags blockers, posts │ ← loop pattern (see below)
│ digests on a schedule │
├─────────────────────────────────────────────────────┤
│ Memory layer Signals from GitHub/Slack/Drive │
│ organized by program, queryable │ ← this repo
│ by any Claude session │
└─────────────────────────────────────────────────────┘
This repo covers the memory layer in full. The proactive layer is documented as a loop pattern — an autonomous Claude agent that wakes up on a schedule, reads program memory, and surfaces what matters without being asked. See content_ideas/serroloop_blog_post.md and the loop pattern concept below. The widget layer remains out of scope.
---
1. Pick your level
Read verdict.md. It has a single table: setup time, what you get, when to move up. Most teams should start at Level 3 (one /loop command). If you're a small org or want the simplest possible start, Level 1 or 2 first.
2. Follow the implementation guide for your level
| Level | File |
|---|---|
| Level 1 — Full Context Pull | [family_a/instructions.md](family_a/instructions.md) |
| Level 2 — Manual Source Mapping | [family_b/instructions.md](family_b/instructions.md) |
| Level 3 — Auto-Ingestion Loop | [family_c/c4_loop.md](family_c/c4_loop.md) (recommended) or [family_c/instructions.md](family_c/instructions.md) for all options |
3. Copy the templates
templates/program_mappings.yaml, templates/charter.md, and templates/CLAUDE_template.md — fill in your org's programs, sources, and owners.
Before you build, read these: - critical_review.md — this repo has a conflict of interest; the review names the biases explicitly - comparative_analysis.md — MCP is pull-only, not a continuous stream; several architectural assumptions fail because of this - family_b/overview.md — Family B has 6 known limitations; long-horizon technical reasoning is the hardest gap to close
---
高质量的MCP工具,实时共享程序内存
该工具使用 NOASSERTION 协议,商用场景请仔细阅读协议条款,必要时咨询法律意见。
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。
📄 NOASSERTION — 请查阅原始协议条款了解具体使用限制。
总体来看,程序内存 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 | program-memory |
| 原始描述 | 开源MCP工具:A starter kit for building live shared program memory using Claude Code and MCP 。⭐10 |
| Topics | mcpclaude codelive shared |
| GitHub | https://github.com/SerroAI/program-memory |
| License | NOASSERTION |
收录时间:2026-06-11 · 更新时间:2026-06-11 · License:NOASSERTION · 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