Mnemoverse is a persistent memory API designed for AI agents, enabling them to store, recall, and learn from preferences, decisions, and lessons across multiple tools using a single API key. The platform addresses the challenge of fragmented or ephemeral memory in AI workflows by providing a unified memory system that works with Claude Code, Claude Desktop, Cursor, VS Code, Windsurf, ChatGPT, Python, and REST interfaces. Unlike vector databases that rely solely on similarity-based retrieval, Mnemoverse incorporates mechanisms inspired by Hebbian associations and prediction-error feedback (Rescorla-Wagner-style), allowing it to learn which memories are most valuable based on outcome feedback and to consolidate or forget information using methods such as HDBSCAN and the Von Restorff effect.
The tool supports writing, learning from, and recalling memories with processes that filter for importance, update memory valence based on reported outcomes, and merge similar memories to maintain a dense, relevant knowledge base. Queries can be made in natural language, with the system expanding results through learned associations and boosting the ranking of memories that have led to positive outcomes. Mnemoverse exposes six core functions—memory_write, memory_read, memory_feedback, memory_stats, memory_delete, and memory_delete_domain—through its @mnemoverse/mcp-memory-server npm package for MCP-compatible clients. The memory graph visualization shows how memories are stored as nodes, linked through associations, and consolidated over time, reflecting the agent’s evolving knowledge.
Developers and teams can integrate Mnemoverse into their AI agents and tools with minimal setup, leveraging its listing on the official MCP Registry and compatibility with popular development environments. The platform has been benchmarked on public memory benchmarks such as LoCoMo and LongMemEval, with all results linked to committed run artifacts for transparency and reproducibility.
Mnemoverse offers a free tier providing 1,000 queries per day, 10,000 atoms, and 60 requests per minute, with no credit card required. Paid plans include Pro and Team tiers with increased limits and additional features such as priority support and shared domains, as well as custom Enterprise options offering unlimited usage, enhanced security, and compliance features. Early-stage startups may be eligible for extended free access to the Team plan.
In the Other AI space, Mnemoverse takes a focused approach. It focuses on allowing AI agents to share and persist memory across different tools and sessions without building custom storage solutions. Mnemoverse is a B2B product aimed at AI developers and agent framework builders. There is a free tier, and paid plans start at $29. Mnemoverse is available on the web, the command line, and API.
Behind Mnemoverse is Mnemoverse, and the product first shipped in 2026. The project is developed in the open on GitHub with 4 commits in the last 90 days. Across PulseGate's embedding index, Mnemoverse has few near neighbours, marking it as relatively distinct. Among its 7 catalogued features are persistent memory, API integration, and MCP server. It exposes integrations via an MCP server and a public API.
Latest indexed changes and source events
mnemoverse.com discovered by the PulseGate indexer
Other apps tracked under the same category.