Enterprise DNA
M MCP Servers Developer low

alibaizhanov/mengram

by Various

Human-like memory for AI agents — semantic, episodic & procedural. Experience-driven procedures that learn from failures. Free API, Python & JS SDKs, LangChain, CrewAI & OpenClaw i

A

MCP

alibaizhanov/mengram

Added 1 June 2026

#agent-memory #ai-agents #ai-memory #claude-desktop #cognitive-architecture #cohere #cursor-ai #episodic-memory

Overview

Mengram is an open-source memory layer for AI agents that implements semantic, episodic, and procedural memory. It learns from failures to refine agent behavior over time and offers a free API with Python and JavaScript SDKs, plus integrations with LangChain, CrewAI, and OpenClaw.

Best for

Best for
Developers building memory-enhanced AI agents who want a free, multi-memory solution with framework integrations

Use cases

  • Give AI agents persistent memory across conversations and sessions
  • Enable agents to learn from past mistakes and improve procedures automatically
  • Augment agentic frameworks like LangChain and CrewAI with structured memory

Notes

Mengram is an open-source memory layer for AI agents that implements semantic, episodic, and procedural memory. It learns from failures to refine agent behavior over time and offers a free API with Python and JavaScript SDKs, plus integrations with LangChain, CrewAI, and OpenClaw.

172 stars on GitHub. Last updated 2026-06-01. Licensed Apache-2.0.

Use cases

  • Give AI agents persistent memory across conversations and sessions
  • Enable agents to learn from past mistakes and improve procedures automatically
  • Augment agentic frameworks like LangChain and CrewAI with structured memory

Pros

  • Free API and open-source code reduce cost and vendor lock-in
  • Covers multiple memory types (semantic, episodic, procedural) in one library
  • Integrates with popular agent frameworks out of the box

Cons

  • Small project with 172 stars may have limited community support
  • Dependency on Python and JS SDKs restricts non-Python/JS stacks
  • Procedural memory learning from failures is experimental and not battle-tested at scale

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

Pros

  • Free API and open-source code reduce cost and vendor lock-in
  • Covers multiple memory types (semantic, episodic, procedural) in one library
  • Integrates with popular agent frameworks out of the box

Cons

  • Small project with 172 stars may have limited community support
  • Dependency on Python and JS SDKs restricts non-Python/JS stacks
  • Procedural memory learning from failures is experimental and not battle-tested at scale

Pairs with

Other entries in the index that connect to this one. Click through to see the chain.