Persistent memory for AI agents — conversation history, context, semantic search, queues. CLI + MCP. As easy as git.
接入配置
选择客户端，复制配置，再按项目文档补齐环境变量。
本地运行
claude mcp add agenticmemory --env AGMRY_API_KEY='<AGMRY_API_KEY>' -- npx -y agmry~/.codex/config.toml
[mcp_servers.agenticmemory]
command = "npx"
args = ["-y", "agmry"]
env = { AGMRY_API_KEY = "<AGMRY_API_KEY>" }
# codex mcp add agenticmemory --env AGMRY_API_KEY='<AGMRY_API_KEY>' -- npx -y agmry~/.cursor/mcp.json
{
"mcpServers": {
"agenticmemory": {
"command": "npx",
"args": [
"-y",
"agmry"
],
"env": {
"AGMRY_API_KEY": "<AGMRY_API_KEY>"
}
}
}
}.vscode/mcp.json
{
"servers": {
"agenticmemory": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"agmry"
],
"env": {
"AGMRY_API_KEY": "<AGMRY_API_KEY>"
}
}
}
}~/.gemini/settings.json
{
"mcpServers": {
"agenticmemory": {
"command": "npx",
"args": [
"-y",
"agmry"
],
"env": {
"AGMRY_API_KEY": "<AGMRY_API_KEY>"
}
}
}
}claude_desktop_config.json
{
"mcpServers": {
"agenticmemory": {
"command": "npx",
"args": [
"-y",
"agmry"
],
"env": {
"AGMRY_API_KEY": "<AGMRY_API_KEY>"
}
}
}
}.trae/mcp.json
{
"mcpServers": {
"agenticmemory": {
"command": "npx",
"args": [
"-y",
"agmry"
],
"env": {
"AGMRY_API_KEY": "<AGMRY_API_KEY>"
}
}
}
}远程托管
claude mcp add --transport sse agenticmemory https://mcp.agenticmemory.ai/sse --header 'Authorization: <AUTHORIZATION>'~/.codex/config.toml
# Codex has no SSE transport: bridged through mcp-remote
[mcp_servers.agenticmemory]
command = "npx"
args = ["-y", "mcp-remote", "https://mcp.agenticmemory.ai/sse", "--header", "Authorization:<AUTHORIZATION>"]~/.cursor/mcp.json
{
"mcpServers": {
"agenticmemory": {
"url": "https://mcp.agenticmemory.ai/sse",
"headers": {
"Authorization": "<AUTHORIZATION>"
}
}
}
}.vscode/mcp.json
{
"servers": {
"agenticmemory": {
"type": "sse",
"url": "https://mcp.agenticmemory.ai/sse",
"headers": {
"Authorization": "<AUTHORIZATION>"
}
}
}
}~/.gemini/settings.json
{
"mcpServers": {
"agenticmemory": {
"url": "https://mcp.agenticmemory.ai/sse",
"headers": {
"Authorization": "<AUTHORIZATION>"
}
}
}
}claude_desktop_config.json · Customize › Connectors › Add custom connector
{
"mcpServers": {
"agenticmemory": {
"command": "npx",
"args": [
"-y",
"mcp-remote",
"https://mcp.agenticmemory.ai/sse",
"--header",
"Authorization:<AUTHORIZATION>"
]
}
}
}.trae/mcp.json
{
"mcpServers": {
"agenticmemory": {
"url": "https://mcp.agenticmemory.ai/sse",
"headers": {
"Authorization": "<AUTHORIZATION>"
}
}
}
}需要的环境变量
AGMRY_API_KEY
把占位值换成你自己的密钥；不要把密钥提交到代码仓库。