Agent Memory
Attach encrypted, agent-owned persistent memory and optional shared knowledge commons to your MCP client.
Overview
agent-memory is a MCP server for the Build phase that provides encrypted persistent agent memory and a shared knowledge commons over MCP.
What is this MCP server?
- Persistent, agent-owned memory with encrypted storage (v0.1.0)
- Shared knowledge commons for cross-agent or team-visible memory patterns
- Hosted SSE remote MCP endpoint on Railway (agent-memory-production)
- GitHub source at MastadoonPrime/agent-memory for self-host exploration
- Focus on long-lived recall rather than ephemeral chat context
- Server version 0.1.0
- Remote transport type SSE
- Open-source repository on GitHub (MastadoonPrime/agent-memory)
Community signal: 1 GitHub stars.
What problem does it solve?
Agents lose bespoke facts and preferences every session, and plain text notes do not give structured, agent-owned recall.
Who is it for?
Indie agent builders experimenting with hosted memory MCP before investing in a custom Supabase or vector stack.
Skip if: Regulated workloads requiring on-prem-only memory with no third-party SSE endpoint, or teams needing mature SLA docs in v0.1.0.
What do I get? / Deliverables
After connecting the SSE MCP endpoint, your agent can read and write encrypted persistent memory and tap shared commons where supported.
- Encrypted persistent memory accessible via MCP tools
- Shared commons layer for reusable agent knowledge
- Remote SSE integration path for quick experiments
Recommended MCP Servers
Journey fit
You install memory MCP while building out how your agent remembers facts, prefs, and project lore—not as a one-off marketing task. The server is purpose-built MCP memory infrastructure (remote SSE), squarely under agent-tooling integrations.
How it compares
Hosted encrypted memory MCP, not an in-repo CLAUDE.md workflow or generic note-taking plugin.
Common Questions / FAQ
Who is agent-memory for?
Solo developers wiring MCP clients who want encrypted, persistent agent memory without building a full backend first.
When should I use agent-memory?
During build-phase agent-tooling setup when session memory limits block multi-day feature work or repeated user preference recall.
How do I add agent-memory to my agent?
Register the remote SSE MCP URL from the server manifest in your agent’s MCP settings, or deploy from the GitHub repo if you need self-hosted control.