Configure persistent shared memory for MCP clients across multiple agents.
npx skills add https://github.com/Lyellr88/MARM-Systems --skill marm-init
Please help me install this Agent Skill. Skill: marm-init Source: https://github.com/Lyellr88/MARM-Systems/tree/main/skills/marm-init Command: npx skills add https://github.com/Lyellr88/MARM-Systems --skill marm-init
This Skill removes the friction of setting up MARM, a persistent local-first memory layer for MCP clients, so agents can share context, retain project decisions, and avoid repeated setup work.
Tell the assistant to guide you through MARM setup for your agent and let it choose the transport, runtime, and connection details.
High-intent search queries and answers about installing and using this skill.
Shared memory for multiple AI agents is configured by validating the MARM engine and writing MCP connection details for each client. This links tools like Claude, Codex, and Cursor to a persistent local-first memory pool so they can read and write shared context.
You can use Docker to run an MCP memory server locally by selecting it as your runtime during setup. The guided installation helps verify the Docker runtime, generate API keys, and confirm server health after launch.
Configuring MCP clients for STDIO or HTTP transport involves choosing the appropriate connection type and writing the corresponding MCP configuration entries. The setup process validates the transport selection and verifies server health across your agents.
Shared agent memory works across different coding tools like VS Code and Cursor by linking MARM across MCP-capable clients. This multi-agent linking allows tools like Gemini, Qwen, and Claude to access the same persistent memory pool.
Accessing a memory dashboard for browsing agent sessions is handled by starting or exposing the web dashboard during setup. This provides an interface to browse retained memories, active sessions, and logs from your connected agents.
Agents lose project context between sessions because they lack a persistent memory layer. Implementing a local-first shared memory infrastructure like MARM allows agents to retain decisions and share context without manual setup work.
RHEcosystemAppEngValidate Lightspeed MCP configuration and runtime connectivity in .mcp.json.
ATTCKDigitalBrowse Smith vault session logs, sub-agent memory, and queue status.
neo4j-contribConfigures the official Neo4j MCP server for schema exploration and Cypher execution.
KorCaptainAutomate MCP server configuration setup with environment checks and verification steps.
