lifeos-mcp
Provides MCP-compatible AI agents with read access to a LifeOS knowledge base, exposing identity, preferences, projects, wiki, skills, and rules via tools and resources.
README
LifeOS MCP
Give any AI agent read access to your LifeOS knowledge base — Claude Desktop, Claude Code, Cursor, ChatGPT, Codex, or any MCP-compatible client.
The server reads from ~/lifeos/ and surfaces everything through lifeos:// resource URIs and lifeos__* tool names.
Install
Once published (coming soon)
npx lifeos-mcp
From source (now)
git clone https://github.com/bwats/lifeos-mcp
cd lifeos-mcp
npm install && npm run build
node bin/lifeos-mcp.js
Global install from source
cd ~/lifeos-mcp
npm install -g .
lifeos-mcp # starts the MCP server on stdio
What it exposes
lifeos-mcp gives any MCP-compatible AI client direct read access to:
| Surface | Path in ~/lifeos/ | Tool |
|---|---|---|
| Identity | identity.md |
lifeos__get_identity |
| Preferences | preferences.md |
lifeos__get_preferences |
| Skills | skills/*/ |
lifeos__list_skills, lifeos__read_skill |
| Projects | projects/*.md |
lifeos__list_projects, lifeos__read_project |
| Wiki | wiki/pages/**/*.md |
lifeos__list_wiki_pages, lifeos__read_wiki_page |
| Rules | system/*.md |
lifeos__list_rules, lifeos__read_rule |
| Any file | (path-safe) | lifeos__read_file |
| Search | (full-text) | lifeos__search |
Modes
| Invocation | What happens |
|---|---|
lifeos-mcp |
Start MCP server on stdio (default) |
lifeos-mcp --stdio |
Start MCP server on stdio (explicit) |
lifeos-mcp list |
List rituals in ~/lifeos/cadence/rituals/ |
lifeos-mcp validate <ritual.md> |
Validate a ritual spec file |
lifeos-mcp render <ritual.md> --to <harness> |
Render ritual to a target harness |
lifeos-mcp --help |
Show all subcommands and options |
Configure MCP clients
Claude Desktop
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
{
"mcpServers": {
"lifeos": {
"command": "npx",
"args": ["lifeos-mcp"]
}
}
}
Or from source:
{
"mcpServers": {
"lifeos": {
"command": "node",
"args": ["/Users/yourname/lifeos-mcp/bin/lifeos-mcp.js"]
}
}
}
Cursor (settings.json)
{
"mcpServers": {
"lifeos": {
"command": "node",
"args": ["/Users/yourname/lifeos-mcp/bin/lifeos-mcp.js"]
}
}
}
Claude Code (~/.claude/settings.json)
{
"mcpServers": {
"lifeos": {
"command": "node",
"args": ["/Users/yourname/lifeos-mcp/bin/lifeos-mcp.js"]
}
}
}
OpenAI Codex (codex.json)
{
"mcpServers": {
"lifeos": {
"command": "node",
"args": ["/Users/yourname/lifeos-mcp/bin/lifeos-mcp.js"]
}
}
}
Custom LifeOS path
Pass --lifeos-path or set LIFEOS_PATH env var:
{
"mcpServers": {
"lifeos": {
"command": "node",
"args": [
"/Users/yourname/lifeos-mcp/bin/lifeos-mcp.js",
"--lifeos-path", "/custom/path/to/lifeos"
]
}
}
}
Environment variables
| Variable | Default | Description |
|---|---|---|
LIFEOS_PATH |
~/lifeos |
Absolute path to the LifeOS root directory |
Tools
| Tool | Description |
|---|---|
lifeos__health |
Check server health and LifeOS availability |
lifeos__get_identity |
Read identity.md — frontmatter + body |
lifeos__get_preferences |
Read preferences.md — frontmatter + body |
lifeos__list_skills |
List all skill directory names in ~/lifeos/skills/ |
lifeos__read_skill |
Read documentation for a named skill |
lifeos__list_rules |
List rule/guideline names from ~/lifeos/system/ |
lifeos__read_rule |
Read a specific rule file |
lifeos__list_projects |
List project filenames from ~/lifeos/projects/ |
lifeos__read_project |
Read a specific project definition |
lifeos__list_wiki_pages |
List all wiki page paths recursively |
lifeos__read_wiki_page |
Read a wiki page by path |
lifeos__search |
Full-text search across all LifeOS files |
lifeos__read_file |
Read any file under ~/lifeos/ by relative path (path-traversal safe) |
lifeos__propose_edge |
Propose a typed semantic edge between two knowledge nodes |
lifeos__accept_edge |
Accept a proposed edge (write-opt-in required) |
lifeos__reject_edge |
Reject a proposed edge (write-opt-in required) |
lifeos__read_pending_queue |
List pending edge proposals (always read-only) |
Resources
| URI | Description |
|---|---|
lifeos://identity |
LifeOS identity profile (identity.md) |
lifeos://preferences |
User preferences (preferences.md) |
lifeos://skills/{name} |
Skill documentation for ~/lifeos/skills/{name}/ |
lifeos://projects/{name} |
Project definition for ~/lifeos/projects/{name}.md |
lifeos://wiki/{+path} |
Wiki page at ~/lifeos/wiki/pages/{path}.md |
lifeos://rules/{name} |
System rule at ~/lifeos/system/{name}.md |
All template resources support listing — clients can enumerate available resources.
CLI subcommands
# List all rituals with id, title, and trigger summary
lifeos-mcp list
# Validate a ritual spec (structural check)
lifeos-mcp validate ~/lifeos/cadence/rituals/morning-review.md
# Render a ritual to a harness
lifeos-mcp render ~/lifeos/cadence/rituals/morning-review.md --to openclaw
lifeos-mcp render ~/lifeos/cadence/rituals/morning-review.md --to claude-code
Renderer interface (pluggable)
Third-party renderers plug in at runtime via registerRenderer:
import { registerRenderer } from "lifeos-mcp/dist/renderers/index.js";
registerRenderer({
name: "my-harness",
async render(spec, body, ctx) {
return {
files: [{ path: "/out/job.json", content: JSON.stringify(spec) }],
summary: `Rendered ${spec.id} to my-harness`,
};
},
});
Built-in renderers: openclaw (cron job JSON), claude-code (slash command + launchd plist).
Security
- All file paths are validated to stay within the LifeOS root — no path traversal possible.
- The server is read-only by default; write tools (
lifeos__propose_edge,lifeos__accept_edge,lifeos__reject_edge) require explicit opt-in via~/Library/Application Support/LifeOS/mcp-write-config.json. - Only files under
~/lifeos/(orLIFEOS_PATH) are accessible.
Build
cd ~/lifeos-mcp
npm install
npm run build
# Output: dist/ + bin/lifeos-mcp.js
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。