CLI: service and health

Use these commands before debugging query, TUI, watcher, MCP, or embedding behavior. Most symptoms start with one of three causes: the service is not reachable, config points at the wrong service, or PostgreSQL/pgvector is not ready.

memory service

The service hosts the HTTP API, web UI, and optional Streamable HTTP MCP route.

memory service run
memory service status
memory service enable
memory service disable
memory service restart-all
memory service ensure-api-token

memory service run is a foreground command. A development cargo run --bin memory service run should not daemonize itself. Packaged installs use systemd on Linux or launchd on macOS for background operation.

Development defaults commonly use 127.0.0.1:4250; packaged installs commonly use 127.0.0.1:4040. Check your config and memory service status instead of assuming a port.

memory status

Use status for a combined project-level view.

memory status --project memory
memory status --project memory --json

Status is the fastest single command for checking service reachability, project config, watcher state, MCP status, and common doctor diagnostics.

memory doctor

Doctor performs a deeper environment inspection.

memory doctor
memory doctor --project memory
memory doctor --json

Run it after config edits, package upgrades, database migrations, token changes, or provider changes.

memory health and memory stats

Use health to confirm the backend responds; use stats to confirm the expected project has data.

memory health
memory stats --project memory

If health is green but stats are empty, the service is probably healthy but the project slug or database is not the one you expected.

memory automation

Automation commands inspect background queues and flush queued work when supported.

memory automation --help

Treat automation flushes as write operations. Capture logs before changing queue state when diagnosing failures.