lfrmonteiro99/memento-mcp

Overview

Memento-MCP is a persistent memory server built on the Model Context Protocol. It stores typed memories with decay scoring to manage relevance over time, and injects context in a token-aware manner to respect LLM token limits. Written in TypeScript, it provides a structured way to give AI agents long-term recall across sessions.

Best for

Best for
Developers needing persistent, token-aware memory for MCP-compatible AI agents

Use cases

• Give an AI assistant long-term memory of user preferences across conversations
• Manage context injection for multi-turn agent workflows within token constraints
• Store and retrieve structured memories with decay-based relevance scoring

Notes

Memento-MCP is a persistent memory server built on the Model Context Protocol. It stores typed memories with decay scoring to manage relevance over time, and injects context in a token-aware manner to respect LLM token limits. Written in TypeScript, it provides a structured way to give AI agents long-term recall across sessions.

1 stars on GitHub. Last updated 2026-06-01. Licensed MIT.

Use cases

• Give an AI assistant long-term memory of user preferences across conversations
• Manage context injection for multi-turn agent workflows within token constraints
• Store and retrieve structured memories with decay-based relevance scoring

Pros

• Typed memories enable structured, machine-parseable storage
• Decay scoring automatically deprioritizes stale information
• Token-aware injection helps avoid exceeding context limits

Cons

• Very low GitHub stars (1) indicate early-stage or unproven adoption
• May lack thorough documentation or community support
• Unclear long-term maintenance given single-digit star count

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.