Mem Agent MCP
MCP server for Dria's fine-tuned memory agent model, providing an Obsidian-like personal memory system for AI agents. Stores and retrieves information as interconnected Markdown files with wikilink navigation, entity extraction, and privacy filtering. Supports connectors for importing from ChatGPT exports, Notion, Nuclino, GitHub, and Google Docs.
Score Breakdown
⚙ Agent Friendliness
🔒 Security
Local-first with privacy filtering. No TLS on local server. External API tokens (GitHub, Google) managed by user. Model inference stays local if configured properly.
⚡ Reliability
Best When
You want a local-first, privacy-aware memory system for AI agents with the ability to import from multiple data sources and query via MCP.
Avoid When
You need a simple key-value memory store, lack GPU hardware for local inference, or need enterprise-grade knowledge management.
Use Cases
- • Giving AI agents persistent personal memory with entity relationships via wikilinks
- • Importing conversation history from ChatGPT, Notion, or GitHub into a searchable knowledge base
- • Running a local memory model with privacy filtering to control what information is exposed
- • Integrating persistent agent memory into Claude Desktop or Claude Code workflows
Not For
- • Large-scale enterprise knowledge management
- • Real-time collaborative editing of memory files
- • Production deployment without GPU/Metal hardware for local inference
- • Non-Python environments
Interface
Authentication
No auth for MCP server itself. GitHub connector uses personal access tokens. Google Docs uses OAuth 2.0 access tokens (short-lived). ChatGPT/Notion/Nuclino use file exports.
Pricing
Free software, but local inference requires GPU/Metal hardware. Remote inference via OpenAI/OpenRouter has per-token costs.
Agent Metadata
Known Gotchas
- ⚠ Requires local LLM inference hardware (macOS Metal or Linux GPU) or paid remote API
- ⚠ LM Studio model name format changed — may need to update from mem-agent-mlx-4bit to mem-agent-mlx@4bit
- ⚠ Google Drive OAuth tokens expire in ~1 hour — not suitable for long-running imports without refresh token handling
- ⚠ ARM64 Linux excludes vLLM by default — must configure remote instance separately
- ⚠ Memory model is specialized (Dria mem-agent) — not a general-purpose LLM
- ⚠ Last pushed Nov 2025 — may not be actively maintained
Alternatives
Full Evaluation Report
Comprehensive deep-dive: security analysis, reliability audit, agent experience review, cost modeling, competitive positioning, and improvement roadmap for Mem Agent MCP.
AI-powered analysis · PDF + markdown · Delivered within 30 minutes
Package Brief
Quick verdict, integration guide, cost projections, gotchas with workarounds, and alternatives comparison.
Delivered within 10 minutes
Score Monitoring
Get alerted when this package's AF, security, or reliability scores change significantly. Stay ahead of regressions.
Continuous monitoring
Scores are editorial opinions as of 2026-03-08.