| name | install-automem |
| description | Install AutoMem locally, on Railway or InstaPods, or against an existing endpoint, then wire MCP clients. |
| license | MIT |
| tags | ["automem","install","mcp","memory"] |
Install AutoMem
Use this skill when a user wants AutoMem installed or connected to an agent,
editor, or MCP-compatible client.
Safety Rules
- Show what it will do before running any command.
- Ask for approval before executing the installer, editing config files, or
provisioning cloud resources.
- Do not print secrets, API keys, bearer tokens, database URLs, or
.env file
contents in chat.
- Treat
https://automem.ai as the marketing and documentation origin. Live API
calls must use the user's deployment-specific AutoMem URL and credentials.
Primary Installer
curl -fsSL get.automem.ai | sh
Setup Flow
- Inspect the user's environment for Node.js, npm, Git, Docker, existing
AutoMem config, and editor/agent config roots.
- Ask which target to use if it is not already clear:
- Local Docker with Qdrant and the local database.
- Railway.
- InstaPods.
- Existing AutoMem endpoint.
- Stage the installer command with the right environment variables, then show
the resulting plan and files that may be written.
- After approval, run the installer, verify
/health, verify authenticated
recall when credentials are available, and wire MCP into the selected clients.
Useful Flags
AUTOMEM_INSTALL_TARGET=local
AUTOMEM_INSTALL_TARGET=cloud
AUTOMEM_INSTALL_TARGET=existing
AUTOMEM_CLIENTS=codex,claude-code,cursor,openclaw,hermes
AUTOMEM_API_URL=https://memory.example
AUTOMEM_DRY_RUN=1
AUTOMEM_NO_AGENT_INSTALL=1
References