CMEM
A memory system for AI coding agents that remembers past decisions, fixes, and dead ends, and can sync that memory across different tools and devices.
🔗 Visit CMEM
Description
AI coding agents forget everything between sessions — the workaround you found for a tricky bug last week, the approach that didn't work, the reasoning behind a decision. CMEM gives agents a persistent memory: it automatically captures decisions, fixes, and dead ends as structured notes, and keeps that memory available whether you switch from Claude Code to Cursor or move to a different machine.
The core, claude-mem, is free and open source (Apache 2.0). CMEM Cloud, the paid layer, adds semantic vector search over that memory, offline sync with a cloud mirror, a private MCP endpoint reachable from any device, and 11 built-in skills with no setup required, at $20/month for individuals or $333/month for teams of 3-50 seats.
💬 Our review
The short version: the free, open-source core (claude-mem) is worth adopting on its own if you want persistent agent memory without paying anything; the $20/month Cloud tier is worth it specifically if you switch between multiple machines or tools and need that memory to follow you.
Against Continue.dev and GitHub Copilot, which focus on in-editor AI assistance rather than cross-session, cross-tool memory persistence, CMEM's specific niche — remembering context across Claude Code, Cursor, Windsurf, and other MCP-compatible clients — isn't really contested by either. The team pricing ($333/month for 3-50 seats) can feel steep for a small startup, and there's real initial setup involved despite the marketing emphasis on simplicity, so budget an afternoon to configure it properly before judging whether it's worth the subscription.
📊 Global score
🤖 AI-enriched data
claude-mem open source gratuit (Apache 2.0). CMEM Cloud : 20 $/mois individuel ou 333 $/mois pour équipes de 3-50 sièges.
Pros
Recherche vectorielle sémantique
Synchronisation hors ligne avec miroir cloud
Endpoint MCP privé
11 compétences intégrées
Récupération sub-seconde
Cons
Configuration initiale nécessaire
Tarif équipe potentiellement élevé
Dépendant de la compatibilité MCP