TencentCloud/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.

What it solves

It addresses the problem of repetitive work and "forgetting" in AI agents. Instead of every agent starting from scratch in every session, this project provides a persistent, shared memory layer that allows agents to reuse project context, documented knowledge, and proven workflows across different sessions and different agent frameworks.

How it works

The system consists of a Memory Core, a Memory Hub (control panel), and a Proxy. It converts raw information into four types of "Memory Assets":

  • Chat Memory: Distills raw conversations into layers (L0 raw $\rightarrow$ L1 atoms $\rightarrow$ L2 scenarios $\rightarrow$ L3 personas) to retain preferences and facts.
  • Skills: Extracts executable workflows, tool calls, and validation rules from successful tasks into reusable modules.
  • Wiki: Converts documents and specs into structured pages with a link graph.
  • CodeGraph: Indexes code symbols and call relationships to allow agents to perform impact analysis before making changes.

Agents connect via a Proxy that requires no code changes to the agent itself. The Memory Hub allows humans to manage these assets, defining which agent has access to which memory (loadouts) and managing visibility (private, team, or restricted).

Who it’s for

  • Developers building teams of specialized AI agents (e.g., a researcher, a coder, and a reviewer) who need to share a common knowledge base.
  • Users of agent frameworks like Claude Code, DeepSeek Harness, or OpenClaw who want their agents to remember context across sessions.
  • Teams wanting to maintain a "save file" of project experience to avoid the cold-start problem for new agents.

Highlights

  • Framework Agnostic: Works with multiple agents via a proxy without needing plugins or MCP servers.
  • Layered Memory: Uses a multi-level distillation process to prevent context window overflow while maintaining detail.
  • Human-in-the-loop: A central panel for reviewing, sharing, and equipping memory assets.
  • Cold Start Capability: Ability to import existing codebases, documents, and chat histories to immediately bootstrap agent knowledge.

Related

  • Project
    mem9-ai/mem9

    A persistent memory layer for AI agents that enables shared context and recall across sessions, machines, and different agent runtimes.

  • Project
    beilusaiying/always-accompany

    A multi-purpose AI agent framework focused on context and attention mechanisms, providing layered memory and dynamic retrieval for long-term companionship, coding, and work.

  • Project
    Fergana-Labs/stash

    A persistent knowledge base and shared memory layer for AI agents that connects to various data sources and records session transcripts to maintain context across runs.

  • Project
    peterfei/ai-agent-team

    A virtual AI development team with specialized roles (PM, Backend, Frontend, etc.) and a persistent memory system for managing complex software projects.

  • Project
    redis/agent-memory-server

    A structured memory layer for AI agents that provides intelligent short-term session memory and persistent long-term context using a two-tier storage model.