Thanks for helping keep this list useful. Recommendations of tools, frameworks, MCP memory servers, benchmarks, and research are all welcome, as are link fixes and better categorization.
- Managed APIs and services: hosted memory services whose primary purpose is memory for LLM/agent applications.
- Open-source frameworks and engines: libraries and engines you run yourself to give agents long-term memory.
- MCP memory servers: memory servers that connect to Model Context Protocol clients.
- Benchmarks and evaluation: suites used to measure agent-memory quality.
- Papers: peer-reviewed or arXiv work on agent memory, long-term context, and memory evaluation.
General-purpose vector databases, RAG frameworks, or agent frameworks where memory is a minor feature are out of scope.
One entry per pull request, placed in the right section, in alphabetical order:
- [Name](https://link) - One factual sentence describing what it does.- Link to a primary source: the project's GitHub repo, official homepage, or the arXiv abstract page.
- Descriptions are factual, not promotional. No superlatives, no pricing, no marketing copy, no addressing the reader.
- Keep it to one line. Name the license and any managed/hosted option when relevant.
- No em dashes in descriptions.
- Check the entry is not already listed (search the README; projects sometimes appear under a different name).
- Verify every link resolves.
- Confirm placement: correct section, correct alphabetical position.
This list includes managed services and their competitors side by side. Entries are described on the same factual terms regardless of who maintains the list. Stars, funding, and popularity are not ranking criteria.
By contributing you agree that your contributions are released under CC0-1.0, like the rest of this list.