Run MCP safely across AI clients
LightNow is the MCP control plane for AI clients: evaluate servers, manage them in Runtime Profiles, and connect clients through LightNow Proxy wherever they run.
Clients stay simple while LightNow resolves profile, policy, secrets, and upstream connections.
Supported clients connect through a LightNow Proxy running alongside them.
One proxy entry eachResolves the active Runtime Profile on the client host and forwards each MCP request to the selected upstream.
Server config and secrets stay managed outside the client file.
Policy checkedStart with your job
Install the local tools, protect useful existing config, connect Codex, and verify the result.
Bring direct MCP entries into a Runtime Profile before LightNow replaces them with one proxy entry.
See the exact files LightNow writes, the safe migration path, and the checks that prove Codex is managed.
Separate client config drift, profile problems, and upstream MCP server failures without exposing secrets.
The safe first path
lightnow login lightnow import-config --client codex --dry-run lightnow import-config --client codex lightnow sync --client codex --local-proxy lightnow config-status --client codex --json lightnow-proxy --config ~/.lightnow/lightnow-proxy/codex.yaml --health --json
Skip both import commands when the Codex config does not contain MCP servers
you need to preserve. The default Runtime Profile is used unless you pass
--profile <name>.
What LightNow manages
- Discover Inspect publisher identity, source, transport, capabilities, and trust signals before using a server.
- Profile Keep each server set in a personal or organization Runtime Profile.
- Connect Run LightNow Proxy alongside each supported AI client.
- Govern Apply organization context and managed-client policy wherever they are configured.
- Observe Use proxy health and metadata-only Runtime Events to investigate failures without storing raw tool arguments or results.
Antigravity