RLM-style MCP server that provides a persistent, sandboxed Python workspace for codebase exploration, keeping file data server-side to reduce LLM context usage and return only intentional summaries
Setup
Choose your client, copy the config, then fill in any required environment variables.
claude mcp add rlm-tools -- uvx rlm-tools~/.codex/config.toml
[mcp_servers.rlm-tools]
command = "uvx"
args = ["rlm-tools"]
# codex mcp add rlm-tools -- uvx rlm-tools~/.cursor/mcp.json
{
"mcpServers": {
"rlm-tools": {
"command": "uvx",
"args": [
"rlm-tools"
]
}
}
}.vscode/mcp.json
{
"servers": {
"rlm-tools": {
"type": "stdio",
"command": "uvx",
"args": [
"rlm-tools"
]
}
}
}~/.gemini/settings.json
{
"mcpServers": {
"rlm-tools": {
"command": "uvx",
"args": [
"rlm-tools"
]
}
}
}claude_desktop_config.json
{
"mcpServers": {
"rlm-tools": {
"command": "uvx",
"args": [
"rlm-tools"
]
}
}
}.trae/mcp.json
{
"mcpServers": {
"rlm-tools": {
"command": "uvx",
"args": [
"rlm-tools"
]
}
}
}