What is an MCP stdio?
MCP stdio is the transport in which a client launches an MCP server as a local subprocess and exchanges JSON-RPC messages with it over the process's standard input and output streams.
Stdio is the simplest MCP transport: no network stack, no ports, no authentication handshake — the host spawns the server binary as a child process, writes newline-delimited JSON-RPC requests to its stdin, and reads responses from its stdout. Because the server runs locally under the host's own process tree, it inherits the host's file-system and OS-level permissions directly, which is exactly right for servers that wrap local resources like a filesystem, a local git repo, or a local database.
Its simplicity is also its main constraint: stdio only works when client and server run on the same machine, since standard streams don't cross a network boundary. There's no built-in authentication layer either, because trust is established implicitly by the fact that the host chose to launch that specific executable; anyone who can get the host to spawn an arbitrary binary as an MCP server effectively controls what that server can do.
For this reason, stdio servers are typically used for trusted, locally-installed integrations, while cross-machine or cloud-hosted integrations use a network transport like Streamable HTTP instead.
In practice with Neotask
Neotask's sandboxed agent runtime launches certain local-capability MCP servers over stdio inside the sandbox container itself, keeping the server's filesystem access naturally scoped to that container's isolated environment rather than exposing it over any network path.
Related terms
- mcp-transport
- mcp-streamable-http
- mcp-server
- mcp-roots
Plans
Free
$0/mo
Download without a card and start for free.
Individual
$50/mo
The full personal agent platform for one person.
Business
$100/mo
One company workspace with room to add your team.
Enterprise
$200/mo
Multiple workspaces and capacity for larger teams.
Continue