{"slug": "tencentdb-agent-memory-a-team-level-memory-hub-for-ai-agents", "title": "TencentDB Agent Memory: A team-level memory hub for AI agents", "summary": "Tencent launched TencentDB Agent Memory, a team-level memory hub for AI agents that automatically extracts and organizes reusable memory assets such as Chat Memory, Skills, Wiki, and CodeGraph, reducing repetitive work and improving efficiency. The open-source project, available on GitHub, supports multi-agent compatibility and cold-start friendly imports, with a one-command deployment script and a data migration tool for existing users.", "body_md": "Latest:Team Memory Beta is evolving quickly — install it and start exploring in minutes.\n\n## memoryhub_demo.mov\n\nStart all three services in one go (`memory-core`\n\n+ `memory-hub`\n\n+ `proxy`\n\n):\n\n```\ngit clone https://github.com/Tencent/TencentDB-Agent-Memory.git\ncd TencentDB-Agent-Memory/deploy/global-images\ncp .env.example .env\n$EDITOR .env       # Fill in two sets of LLM parameters (memory group + proxy group)\n./start-all.sh     # Launch everything with one command; when finished, it prints a one-liner you can paste directly into Claude\n```\n\nOpen the panel: [http://localhost:8125](http://localhost:8125).\n\nComplete installation documentation (standalone Memory Hub deployment, Proxy + Claude Code / CodeBuddy usage, stop and cleanup, port reference, etc.) is available in [ INSTALL.md](/TencentCloud/TencentDB-Agent-Memory/blob/feat/server_team/INSTALL.md) (中文:\n\n[INSTALL_CN.md](/TencentCloud/TencentDB-Agent-Memory/blob/feat/server_team/INSTALL_CN.md)).\n\nIf you're already on an older release (v1.x / v0.x) and want to bring your existing data over to v2.0.0+, we provide a migration tool:\n\nSee [ Data Migration Tool (v2 → v3)](/TencentCloud/TencentDB-Agent-Memory/blob/feat/server_team/MemoryCore/scripts/migrate-v2-to-v3/README.md) for full usage and flags. New installations can skip this.\n\nWe started from a practical question: **How do you reduce repetitive work when using Agents?**\n\nIf project context has already been explained, it shouldn't need to be repeated in a new session. If documents have already been read, every Agent shouldn't have to start again from page one. A workflow that already works shouldn't have to be rediscovered next time.\n\nMemory here means more than just \"remembering conversations.\" **Any information that helps the next Agent avoid reinventing the wheel should be saved, organized, and reused.**\n\n```\nExisting information → Reusable memory assets → Fewer turns → Less rework → More stable results and higher efficiency\n```\n\n**Memory Hub** for Agent teams closes the loop across the entire experience lifecycle: work produces assets, assets circulate through the team, and new members can load the team's save file on day one.\n\n**Automatic asset extraction**: Extract Chat Memory and Skills from conversations and tasks; convert documents and code into Wiki and CodeGraph; then manage, review, and route them consistently.**Portable & multi-Agent compatible**: Memory assets are decoupled from Agent frameworks — they can move across frameworks and be shared and maintained by multiple Agents and team members.**Cold-start friendly**: Import existing documents, codebases, and Agent conversation sessions. New Agent teams can start from existing experience instead of learning from scratch.\n\n**Chat Memory** retains preferences, facts, decisions, and interaction history.- Each Agent automatically gets its own memory when created — no need to re-introduce yourself next time.\n- L0 Conversation → L1 Atom → L2 Scenario → L3 Persona — raw conversations are distilled layer by layer.\n\n\"Don't refactor the old auth module — mobile is still using it.\" — Context this costly shouldn't depend on humans repeating it every time.\n\n- After completing complex work, Agents can extract and manage reusable Skills from conversations and tool calls, and import them into the context of a designated Agent when needed.\n- A Skill isn't just a prompt snippet; it has versions, resource files, trigger boundaries, execution steps, and validation rules.\n- Personal Skills are private by default; after review, they can be shared with the team and assigned to other Agents.\n\nTroubleshooting, code review, release checklists — learn it once, and the whole team can use it.\n\n**Wiki** turns product docs, design specs, and ops runbooks into structured pages with a link graph. (Inspired by Karpathy's LLM knowledge base.)\n\n**CodeGraph** indexes code symbols, files, call relationships, and impact paths.\n\n- Agents can search, read, inspect callers/callees, and perform impact analysis before modifying code.\n\nWiki keeps Agents from reading every file list before getting to work. CodeGraph doesn't just tell them \"the code is here\" — it tells them \"changing this might affect those.\"\n\n- Create teams and Agents in Memory Hub; review, share, and equip memory assets.\n- Manage ownership, versions, status, visibility, usage counts, and Agent bindings in one place.\n`private`\n\nbelongs strictly to the Owner;`team`\n\nis visible to all team members;`restricted`\n\ngrants precise access via User / Role / Agent ACLs.- Two role layers:\n**global System Admin** manages users and teams (creating teams, adding members) and can also use Wiki, CodeGraph, Skill, and other asset management features;**Team-level roles** include Admin (team manager) and Member (regular member), responsible for asset collaboration and access control within a team. Asset ownership is tracked via Owner — the Owner automatically has management permissions for their assets.\n\nMost Agents' first task is re-learning your project. TencentDB Agent Memory turns the learning cost you've already paid into a save file:\n\nSpecifically, these existing assets can be imported directly and processed automatically in the panel:\n\n**Codebases**: Import existing repositories —** CodeGraph**automatically indexes symbols, files, call relationships, and impact paths.** Documents & files**: Import relevant docs and files —** Wiki**automatically generates structured pages with a link graph.** Conversation sessions**: Import past Agent conversation sessions —** Skills and Chat Memory**are automatically extracted as reusable assets.\n\nStop retraining every Agent. Give it the save file.\n\nOpen Memory Hub and create a team:\n\n```\nTiny but Serious Inc.\n├── 👤 You · Set goals / Make decisions\n├── 🔭 Scout · Research / Find opportunities\n├── 🛠 Builder · Write code / Build products\n├── 🧪 Reviewer · Test / Find issues\n└── 🧠 Agent Memory · Preserve the team's experience\n```\n\nYou're not opening four disconnected chat windows — you're assembling a squad with different roles that can inherit the team's accumulated experience.\n\n```\n🔭 Scout\n   ├── User interview Chat Memory\n   ├── Market research Wiki\n   └── Competitive analysis Skill\n\n🛠 Builder\n   ├── Product Wiki\n   ├── Project CodeGraph\n   └── Feature Delivery Skill\n\n🧪 Reviewer\n   ├── Historical incident Chat Memory\n   ├── Project CodeGraph\n   └── Release Checklist Skill\n```\n\nDifferent roles, different loadouts. Less noise — give each Agent the memory assets it actually needs to get work done.\n\n**The company can be tiny. Experience can compound forever.**\n\nRAG answers \"what can be found?\" Team Memory also answers \"who can use it, which version is valid, and which Agent should receive it.\"\n\n| Chat History | Standard RAG | TencentDB Agent Memory | |\n|---|---|---|---|\n| Cross-session user understanding | △ | △ | ✅ Chat Memory |\n| Distilled executable experience | — | — | ✅ Skill |\n| Document structure & relationships | — | △ Chunk retrieval | ✅ Wiki + Link Graph |\n| Code call graphs & impact scope | — | △ Text match | ✅ CodeGraph |\n| Ownership / Version / Status | — | — | ✅ |\n| Team sharing & Agent loadout | — | — | ✅ |\n| Private / Team / ACL | — | △ | ✅ |\n\n| Play Style | What you do in the Hub |\n|---|---|\nTeam Up |\nCreate teams, add people and Agents, define sharing boundaries |\nAsset Library |\nBrowse, search, review, and manage Chat Memory, Skills, Wiki, and CodeGraph |\nAgent Loadout |\nBind different memory assets to different Agents; adjust priority and usage mode |\nKnowledge Workshop |\nBuild Wiki and CodeGraph; monitor processing status and asset metadata |\nAccess Control |\nSwitch between private, team, and ACL-based access; revoke sharing when needed |\n\nWhen you open an asset, what matters is not just \"what it says,\" but also \"where it came from, which version it is, who it's assigned to, and whether it's been used recently.\"\n\nMemory doesn't run the Agent loop; it ensures the next iteration inherits the previous one's results: valuable interactions stay in Chat Memory, proven workflows are distilled into Skills, and document/code changes are updated through Wiki ingest and CodeGraph sync.\n\n**Without Memory, loops may just repeat faster. With inherited memory, each iteration has the chance to be better than the last.**\n\nNew Chat Memory and Skills are private by default. Sharing is an explicit action, not a default leak.\n\n| Visibility | Semantics |\n|---|---|\n`private` |\nOnly the Owner can read — not even team admins |\n`team` |\nTeam members can read; the Owner / Admin can manage |\n`restricted` |\nPrecise access via User / Role / Agent ACL |\n`agent` |\nFor targeted equipping of Agents within the same team |\n\nYou can assign the \"Release Skill\" to the Release Agent, the \"Architecture Wiki\" to all development Agents, and CodeGraph to Coder and Reviewer.\n\nTencentDB Agent Memory doesn't aim to \"store everything.\" It solves three problems: **what's worth keeping, who can use it, and how to retrieve less while retrieving the right things next time.**\n\nConversations are first saved as L0, then refined by an async pipeline into multiple levels of granularity:\n\n| Layer | What it stores | Primary use |\n|---|---|---|\nL0 Conversation |\nRaw conversations with full context | Verify exact wording, timestamps, and sources |\nL1 Atom |\nFacts, preferences, constraints, and events extracted from conversations | Precise recall of actionable information |\nL2 Scenario |\nKnowledge blocks organized around projects or scenarios | Quickly restore a working context |\nL3 Core / Persona |\nLong-term profiles, stable patterns, and high-level cognition | Let Agents rapidly enter a user's and team's context |\n\nBoth generation and retrieval are layered: normally, L2/L3 provide a quick context bootstrap; when specific facts are needed, BM25 + vector retrieval + RRF fall back to L1/L0. Results are further capped by item count, character budget, and timeout limits to prevent memory from overwhelming the context window.\n\nChat Memory, Skills, Wiki, and CodeGraph are all registered uniformly as Memory Assets. Memory Hub uses **Fixed Binding + ACL** to determine which assets a given Agent can use: first narrow the permission scope by Team, User, Agent, and visibility, then retrieve based on the current query.\n\nThis lets teams share experience without exposing all their private information; switching Agents or frameworks only requires re-equipping, not retraining.\n\nDocuments are organized into searchable Wiki pages that support link-graph drill-down; codebases are indexed into CodeGraph assets containing files, symbols, and call relationships. Agents first discover capabilities via `/v3/tools/list`\n\n, then use `/v3/tools/call`\n\nto read relevant pages, source code, or impact paths.\n\nThis makes documents and code part of memory as well — but they remain available tools that only enter context when truly needed.\n\n| Benchmark | Without TencentDB Agent Memory | With it enabled | Relative improvement |\n|---|---|---|---|\nPersonaMem |\n48% | 76% |\n+59% |\n\nPersonaMem tests whether an Agent can correctly understand and apply user information after extended interactions.\n\n- Wiki and CodeGraph are built asynchronously; allow some processing time before they reach\n`ready`\n\nstatus. - CodeGraph currently prioritizes public HTTPS repositories; support for private repositories and SSH credentials is still being refined.\n- The Hub supports manual asset binding; fully automated memory routing is still under iteration.\n- TencentDB Agent Memory currently supports OpenClaw, Hermes, Claude Code, CodeBuddy, and SDK integration; broader cross-framework migration is on the roadmap.\n\n[Full Installation Guide](/TencentCloud/TencentDB-Agent-Memory/blob/feat/server_team/INSTALL.md)(Memory Core + Hub + Proxy one-click deployment)[Data Migration Tool (v2 → v3)](/TencentCloud/TencentDB-Agent-Memory/blob/feat/server_team/MemoryCore/scripts/migrate-v2-to-v3/README.md)(if you're on an older release and want to migrate existing data)[Knowledge OpenAPI](/TencentCloud/TencentDB-Agent-Memory/blob/feat/server_team/MemoryKnowledge/openapi.yaml)[Contributing Guide](/TencentCloud/TencentDB-Agent-Memory/blob/feat/server_team/CONTRIBUTING.md)\n\nAgent Memory doesn't have a settled standard yet. Bug reports, documentation, benchmarks, new framework adapters, and more creative Memory Hub use cases are all welcome.\n\nTencentDB Agent Memory stands on the shoulders of the open-source community:\n\n— our CodeGraph asset module**CodeGraph****uses code from this project**. Its design of a pre-indexed code graph is the foundation of our implementation.(Nous Research) — our Skill asset management**Hermes Agent****uses part of the Skill-related code from Hermes Agent and builds further optimizations base on it**.— the idea of treating documentation as an LLM-maintained, incrementally growing knowledge artifact directly informed how our Wiki layer is built and kept up to date.**\"LLM Wiki\"** by Andrej Karpathy\n\nWe are grateful to the authors and contributors of these projects.\n\nWe welcome contributions of all kinds — bug reports, feature suggestions, documentation fixes, benchmark reproductions, ecosystem integrations, or pull requests. Agent memory is far from settled, and we hope to build it together with the community.\n\n- 🐞\n**Found a bug or have a question?** Open an issue in[GitHub Issues](https://github.com/Tencent/TencentDB-Agent-Memory/issues)— we respond within 24 hours. - 💡\n**Have an idea to share?** Start a thread in[GitHub Discussions](https://github.com/Tencent/TencentDB-Agent-Memory/discussions). - 🛠️\n**Want to contribute code?** Please read[CONTRIBUTING.md](/TencentCloud/TencentDB-Agent-Memory/blob/feat/server_team/CONTRIBUTING.md)first. - 💬\n**Want to chat with us?** Join our[Discord community](https://discord.gg/dJQM6mKMF)and talk to the core developers directly.\n\nLet the path the team has walked become the next Agent's starting line.\n\n💡 Thanks to the following contributors building with us — you make TencentDB Agent Memory better.\n\nIf TencentDB Agent Memory has been helpful to you, please consider starring the project.If you have any suggestions, feel free to open an issue for discussion. |\n|\n\n[MIT](/TencentCloud/TencentDB-Agent-Memory/blob/feat/server_team/LICENSE) © TencentDB Agent Memory Team", "url": "https://wpnews.pro/news/tencentdb-agent-memory-a-team-level-memory-hub-for-ai-agents", "canonical_source": "https://github.com/TencentCloud/TencentDB-Agent-Memory", "published_at": "2026-08-05 15:20:39+00:00", "updated_at": "2026-08-05 15:37:23.337425+00:00", "lang": "en", "topics": ["ai-agents", "ai-tools", "ai-infrastructure"], "entities": ["Tencent", "TencentDB Agent Memory", "Claude", "CodeBuddy", "GitHub"], "alternates": {"html": "https://wpnews.pro/news/tencentdb-agent-memory-a-team-level-memory-hub-for-ai-agents", "markdown": "https://wpnews.pro/news/tencentdb-agent-memory-a-team-level-memory-hub-for-ai-agents.md", "text": "https://wpnews.pro/news/tencentdb-agent-memory-a-team-level-memory-hub-for-ai-agents.txt", "jsonld": "https://wpnews.pro/news/tencentdb-agent-memory-a-team-level-memory-hub-for-ai-agents.jsonld"}}