Skip to content
HUIAHUB

TencentCloud/TencentDB-Agent-Memory

TencentDB-Agent-Memory

TencentDB Agent Memory is a team-level memory hub for AI Agents — turning conversations, docs, and code into four reusable memory assets (Chat Memory, Skill, LLM-Wiki, Code-Graph) that are governed, shared, and equipped across agents and frameworks.

Metrics updated Established
View on GitHub (opens in a new tab)Homepage / demo

GitHub metrics

Stars
28K
growth after 7 days of data
Forks
2.7K
30-day growth
—
Open issues
903

The chart appears once the repository has been checked on two or more days.

What it does

TencentDB Agent Memory is a self-hosted memory service from Tencent Cloud that stores what AI agents learn and shares it across agents and team members.

It turns past work into four kinds of memory asset. Chat Memory keeps preferences, facts and decisions, distilled from raw conversation through layers named L0 Conversation, L1 Atom, L2 Scenario and L3 Persona. Skills capture reusable procedures with versions and validation rules. LLM-Wiki builds linked pages from documents, and CodeGraph indexes symbols, files and call relationships so an agent can check what a change might affect. A web panel called Memory Hub tracks ownership, versions and access, with private, team and ACL-based visibility.

Agents connect through a proxy. You point the agent's base URL at it, and the README says no plugin, hook or MCP server is needed. Listed clients include Claude Code, Codex, CodeBuddy, WorkBuddy, Hermes and OpenClaw. Setup starts three services (memory-core, memory-hub and proxy) with Docker and a start script, and asks for two sets of LLM settings. The README reports a PersonaMem score of 76% with the system against 48% without it, which is the project's own figure. Team Memory is labeled beta.

Why it matters

Agents usually start each session without the context earlier sessions built up. This project stores that context as named, versioned assets with owners and permissions, so a team decides which agent receives which memory instead of sharing one global prompt. Integration happens at the proxy, so a single memory store can serve several agent tools at once without code changes in each of them.

Main features

  • Four memory asset types: Chat Memory, Skill, LLM-Wiki and Code-Graph
  • Layered conversation memory from L0 Conversation up to L3 Persona
  • Retrieval that combines BM25, vector search and RRF
  • Proxy integration by changing an agent's base URL
  • Memory Hub panel at http://localhost:8125 for teams, assets and bindings
  • Visibility levels: private, team, restricted and agent
  • Imports codebases, documents and past agent sessions for a cold start
  • Migration tool for data from v1.x and v0.x releases

Who it's for

  • Teams running several coding or work agents that should share context
  • Claude Code, Codex or OpenClaw users who want memory that persists across sessions
  • Organizations that need ownership and access control over agent memory
  • Developers comfortable running Docker services and configuring LLM endpoints

Strengths

  • One proxy serves seven named agent clients plus a generic integration path.
  • New Chat Memory and Skills are private by default, and sharing is an explicit action.
  • Covers documents and code as well as chat history.
  • The repository's LICENSE file states MIT terms.

Limitations

  • Team Memory is labeled beta and is changing quickly.
  • The MongoDB storage backend is experimental and off by default.
  • CodeGraph favors public HTTPS repositories, and private repos and SSH credentials are still being refined.
  • Binding assets to agents is manual, and automated memory routing is still in development.

Getting started

Requirements: Docker, git and two sets of LLM endpoint settings (a memory group and a proxy group); the README badge lists Node >=22.16.

Always follow the official instructions in the README on GitHub. This overview is our summary, not a copy.

Sources