OpenAI-compatible proxy that builds a knowledge graph of tasks, decisions and files in a session and replays it when the context window runs out.
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。
claude mcp add tokenmizer --env TOKENMIZER_API_KEY='<TOKENMIZER_API_KEY>' -- uvx tokenmizer~/.codex/config.toml
[mcp_servers.tokenmizer]
command = "uvx"
args = ["tokenmizer"]
env = { TOKENMIZER_API_KEY = "<TOKENMIZER_API_KEY>" }
# codex mcp add tokenmizer --env TOKENMIZER_API_KEY='<TOKENMIZER_API_KEY>' -- uvx tokenmizer~/.cursor/mcp.json
{
"mcpServers": {
"tokenmizer": {
"command": "uvx",
"args": [
"tokenmizer"
],
"env": {
"TOKENMIZER_API_KEY": "<TOKENMIZER_API_KEY>"
}
}
}
}.vscode/mcp.json
{
"servers": {
"tokenmizer": {
"type": "stdio",
"command": "uvx",
"args": [
"tokenmizer"
],
"env": {
"TOKENMIZER_API_KEY": "<TOKENMIZER_API_KEY>"
}
}
}
}~/.gemini/settings.json
{
"mcpServers": {
"tokenmizer": {
"command": "uvx",
"args": [
"tokenmizer"
],
"env": {
"TOKENMIZER_API_KEY": "<TOKENMIZER_API_KEY>"
}
}
}
}claude_desktop_config.json
{
"mcpServers": {
"tokenmizer": {
"command": "uvx",
"args": [
"tokenmizer"
],
"env": {
"TOKENMIZER_API_KEY": "<TOKENMIZER_API_KEY>"
}
}
}
}.trae/mcp.json
{
"mcpServers": {
"tokenmizer": {
"command": "uvx",
"args": [
"tokenmizer"
],
"env": {
"TOKENMIZER_API_KEY": "<TOKENMIZER_API_KEY>"
}
}
}
}需要的环境变量
TOKENMIZER_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。