Skip to main content

​
memanto connect

Install Memanto into a supported AI coding tool, IDE, or agent. Each target gets a tool-specific instruction file (e.g. CLAUDE.md, .cursor/rules/memanto.mdc) and the memanto-memory skill. Both files are stamped with that target’s own slug, so the connected tool identifies itself on every memory call — see Tool attribution.
Arguments:
  • TARGET - One of the supported target IDs below.
Supported Targets: Options:
  • -p, --project-dir TEXT - Target project directory (default: .)
  • -g, --global - Install globally to the target’s home directory (e.g. ~/.claude/, ~/.cursor/) instead of the current project
Examples: Install for the current project:
Install for a different project:
Install globally (system-wide):
Output (example):
Notes:
  • Run memanto connect list to see which targets are detected in your project and which already have Memanto installed.
  • Use memanto connect multi for an interactive multi-target setup.
  • Use memanto connect remove <target> to disconnect.
  • Claude Code and Grok Build can install the Memanto plugin from the plugin marketplace instead. It ships the same hooks plus slash commands, a memory-scout subagent, and a status line. Use either the plugin or memanto connect claude-code, not both.
  • Some targets get extra artifacts that run memanto memory sync automatically, keeping the dynamic memories in their instruction file current: claude-code, codex, and cursor get hooks written into their own settings file, and pi gets a memanto-sync.ts extension. Claude Code additionally gets a PostToolUse hook that summarizes memanto commands in the transcript.
  • Instruction files are stamped with a template version. After upgrading the memanto package, run memanto connect update to refresh already-connected targets.