ReMe vs TencentDB-Agent-Memory
Persistent AI agent memory projects compared
ReMe and TencentDB-Agent-Memory turn conversations and other resources into reusable memory for AI agents. ReMe focuses on a local, editable Markdown workspace, while TencentDB-Agent-Memory emphasizes shared team assets, access controls, and agent bindings.
ReMe: Manage Persistent Memory for AI Agents
ReMe turns agent conversations and resources into searchable, editable Markdown memory that can evolve over time. It suits developers building assistants that need reusable knowledge across sessions and agent runtimes.
TencentDB-Agent-Memory: Share Reusable Memory Across AI Agents
TencentDB Agent Memory is a team memory hub that turns conversations, documents, and code into reusable assets for AI agents. It suits teams that want agents to share curated context and skills across sessions and frameworks.
| ReMe | TencentDB-Agent-Memory | |
|---|---|---|
| Language | Python | TypeScript |
| License | Apache-2.0 | NOASSERTION |
| Stars | 3.5k | 27.7k |
| Forks | 306 | 2.7k |
| Last analyzed | Oct 3, 2026 | Oct 3, 2026 |
Key differences
- ReMe stores memories as editable Markdown with frontmatter and wikilinks; TencentDB-Agent-Memory organizes layered chat memory, skills, a linked wiki, and code indexes.
- ReMe offers CLI, HTTP, MCP, and Python interfaces; TencentDB-Agent-Memory provides a proxy for supported agent clients and a hub for managing and assigning assets.
- ReMe is written in Python and lists an Apache-2.0 license; TencentDB-Agent-Memory is written in TypeScript, and its repository reports NOASSERTION while its README displays an MIT badge.
- ReMe’s default setup is a local service and workspace; TencentDB-Agent-Memory is a self-hostable hub and proxy, with documented Docker deployment.
- ReMe suits individual or multi-runtime use where people want control over memory files; TencentDB-Agent-Memory focuses on teams sharing assets with visibility controls and agent access rules.
- The supplied repository data lists 3.5k stars for ReMe and 27.7k for TencentDB-Agent-Memory. Both are marked as not archived.
Choose ReMe if you…
- want durable, human-readable Markdown memories that you can inspect and edit.
- need CLI, HTTP, MCP, or Python access to a local memory workspace.
- prefer BM25 and wikilink retrieval, with optional embedding-based search.
Choose TencentDB-Agent-Memory if you…
- need people and multiple agents to share curated project knowledge with access controls.
- want reusable Skills, document-derived wiki content, and code relationship searches in one hub.
- plan to seed team memory by importing existing repositories, documents, or conversation sessions.
This comparison is generated with AI from the OSRepos analyses of both projects. Always check each project's repository and documentation before choosing.