Flisk as a Control Plane: Let Your Coding Assistant Run the Whole Local Stack
The mcp resource puts an agent's tools inside your stack. The flisk mcp command does the reverse: it lets an assistant control the stack. Run it over stdio, and a coding assistant can check the machine, start services, read their status, and read bounded logs. Connect it The Flisk docs give the same entry for the MCP clients they name, Claude Code, Cursor, and Codex: { "mcpServers" : { "flisk" :…
The Flisk MCP (Machine Control Plane) command allows a coding assistant to manage and control a local development stack. By interacting through standard input/output, the assistant can perform various tasks such as checking the machine's status, starting services, reading service statuses, and fetching bounded logs for debugging purposes.
To utilize this MCP, you must first set up the mcpServers configuration in the Flisk documentation, specifying the flisk command with the mcp argument. This server should be executed from the project folder to ensure it can locate the project's configuration. The assistant can perform four primary operations: doctor, which diagnoses container runtime and port conflicts; up, which starts services; down, which stops services; status, which provides structured health information about APIs and databases; and logs, which returns a limited window of output for debugging.
The bounded logs are crucial, as a full log dump can overwhelm the assistant's context and obscure critical information. To effectively manage the local stack with an assistant, follow a structured procedure: run the doctor check, stop and report if there are any runtime issues, start the necessary services, and continuously poll the status until all services are ready or a failure occurs.
In case of failure, read the logs to identify the problematic service, and report the findings along with the status output as evidence. It is important to restrict the assistant's access to sensitive operations such as deleting volumes, which contains project data, and modifying generated files. These tasks should be handled manually to prevent accidental data loss or overwriting of critical project components.
Before delegating the entire stack management to the assistant, it is advisable to test the loop yourself by executing flisk doctor, flisk up, flisk status --json, and flisk down commands. If these commands fail, the assistant will encounter the same errors, but they may be harder to diagnose in the assistant's output.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.