Skip to content

Configuration Reference

l0-memory can be configured via environment variables and VS Code settings.

Environment Variables

These variables affect the behavior of the ltm binary, whether run as a CLI or an MCP server.

VariableDescriptionDefault
LTM_DBAbsolute path to the SQLite database file.~/.long-term-memory/memories.db
LTM_SCOPESets the default scope for all commands.user
LTM_DEBUGIf set to 1, enables verbose debug logging to stderr.(empty)
LTM_LOG_FILEPath to a file where debug logs should be appended.(empty)

Debugging

Setting LTM_LOG_FILE automatically enables LTM_DEBUG. This is useful for troubleshooting MCP connection issues where stderr might be swallowed by the host.

VS Code Settings

These settings are available in the VS Code Settings UI under the l0-memory section.

l0-memory.binaryPath

Type: string | Default: "" The absolute path to the ltm binary. If left empty, the extension will attempt to auto-discover the binary in common locations and the extension's bundled bin/ folder.

l0-memory.dbPath

Type: string | Default: "" Overrides the SQLite database path. This effectively sets the LTM_DB environment variable for the extension's internal operations.

l0-memory.defaultScope

Type: enum | Default: "user" Options: user, ask, repo:current. Determines the scope used when adding a new memory from the sidebar. repo:current will attempt to use the name of your active workspace folder.

l0-memory.groupByScope

Type: boolean | Default: false When enabled, the Memories tree view will group entries under collapsible scope nodes.

l0-memory.sortBy

Type: enum | Default: "updated" Options: updated, created, key, scope. Sets the default sort order for the Memories tree view. Note that pinned items always appear at the top regardless of this setting.

l0-memory.autoStartMCP

Type: boolean | Default: false If true, the extension will spawn a background MCP server on activation. This is generally unnecessary if you are using an MCP host like Claude Code that starts its own instance.

Crafted with precision for AI assistants.