{"slug": "tencentdb-agent-memory-a-team-level-memory-hub-not-just-per-agent-recall", "title": "TencentDB Agent Memory: A Team-Level Memory Hub, Not Just Per-Agent Recall", "summary": "Tencent Cloud released TencentDB Agent Memory, an MIT-licensed memory hub for AI agent teams that converts conversations, documents, and code into four shareable memory assets—Chat Memory, Skill, Wiki, and CodeGraph—using a layered L0-L3 distillation pipeline. The project, which has over 10,600 GitHub stars and is labeled 'Team Memory Beta,' reports a +59% PersonaMem score improvement, addressing the problem of repetitive work in agent workflows by enabling team-level memory sharing rather than per-agent recall.", "body_md": "# TencentDB Agent Memory: A Team-Level Memory Hub, Not Just Per-Agent Recall\n\nTencentDB Agent Memory is Tencent's MIT-licensed memory hub for AI agent teams, turning conversations, docs, and code into four governed, shareable memory assets — Chat Memory, Skill, Wiki, and CodeGraph — with a layered L0-L3 distillation pipeline and a reported +59% PersonaMem score.\n\n- ⭐ 10663\n- Node.js\n- MIT\n- Updated 2026-08-02\n\n[AI Agent Memory: Letta vs Mem0 vs A-Mem](https://dibi8.com/resources/llm-frameworks/ai-agent-memory-persistence-letta-mem0-a-mem-2026/) •\n[Hermes Agent: Self-Improving AI Agent](https://dibi8.com/resources/llm-frameworks/hermes-agent-self-improving-ai-agent/)\n\n*Project banner — from github.com/TencentCloud/TencentDB-Agent-Memory*\n\n## What Is TencentDB Agent Memory? [#](#what-is-tencentdb-agent-memory)\n\n**TencentDB Agent Memory** starts from a specific question the maintainers state directly: *“How do you reduce repetitive work when using Agents?”* If project context has already been explained once, a new session shouldn’t need it re-explained. If a document’s already been read, the next agent shouldn’t start from page one. TencentDB Agent Memory’s answer is a **Memory Hub** that extracts, governs, and routes four types of reusable memory assets across a *team* of agents — not just a single agent’s own conversation history.\n\n🔗 **GitHub**: [https://github.com/TencentCloud/TencentDB-Agent-Memory](https://github.com/TencentCloud/TencentDB-Agent-Memory)\n\nMIT licensed, at **10,600+ GitHub stars**, with a commit from **July 29, 2026**, it’s a Tencent Cloud project still explicitly labeled “Team Memory Beta” — evolving quickly rather than a finished, stable product.\n\n## Four Memory Assets, Not One Chat Log [#](#four-memory-assets-not-one-chat-log)\n\n| Chat History | Standard RAG | TencentDB Agent Memory | |\n|---|---|---|---|\n| Cross-session user understanding | Partial | Partial | Chat Memory |\n| Distilled executable experience | No | No | Skill |\n| Document structure & relationships | No | Chunk retrieval only | Wiki + Link Graph |\n| Code call graphs & impact scope | No | Text match only | CodeGraph |\n| Ownership / Version / Status | No | No | Yes |\n| Team sharing & Agent loadout | No | No | Yes |\n| Private / Team / ACL | No | Partial | Yes |\n\nThe framing the maintainers draw: RAG answers “what can be found?” — TencentDB Agent Memory also answers **“who can use it, which version is valid, and which Agent should receive it.”**\n\n### The four asset types [#](#the-four-asset-types)\n\n**🧠 Chat Memory**— preferences, facts, decisions, and interaction history. Each agent gets its own automatically on creation. Distilled layer by layer: L0 Conversation → L1 Atom → L2 Scenario → L3 Persona.**⚡ Skill**— a reusable procedure extracted from completed work, with versions, resource files, trigger boundaries, execution steps, and validation rules — private by default, shareable with the team after review.**📖 Wiki**— documents, specs, and runbooks turned into structured, link-graphed pages, explicitly inspired by[Andrej Karpathy’s “LLM Wiki” concept](https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f).**🕸️ CodeGraph**— indexes code symbols, files, call relationships, and impact paths, so an agent can check callers/callees and impact scope before modifying code.\n\n## Installation [#](#installation)\n\n```\ngit clone https://github.com/TencentCloud/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     # Launches memory-core + memory-hub + proxy in one command\n```\n\n`start-all.sh`\n\nstarts all three services (`memory-core`\n\n, `memory-hub`\n\n, `proxy`\n\n) together and prints a one-liner you can paste directly into Claude when it finishes. The panel is then reachable at `http://localhost:8125`\n\n.\n\nMigrating from an older v1.x/v0.x install has a dedicated tool (v2 → v3); new installs can skip it.\n\n## Cold Start: Import What You Already Have [#](#cold-start-import-what-you-already-have)\n\nRather than starting a new agent team from zero, existing assets can be imported directly:\n\n*Cold-start import flow — from github.com/TencentCloud/TencentDB-Agent-Memory*\n\n**Codebases**→ CodeGraph automatically indexes symbols, files, call relationships, and impact paths** Documents & files**→ Wiki automatically generates structured, link-graphed pages** Conversation sessions**→ Skills and Chat Memory are automatically extracted as reusable assets\n\n## Team Play: Building an Agent Team, Not Four Chat Windows [#](#team-play-building-an-agent-team-not-four-chat-windows)\n\nThe README’s own worked example is a “one-person company” with role-specific agents:\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\nEach role gets a different **loadout** of memory assets — not everything, just what that role needs:\n\n```\nScout:     User-interview Chat Memory, Market-research Wiki, Competitive-analysis Skill\nBuilder:   Product Wiki, Project CodeGraph, Feature-Delivery Skill\nReviewer:  Historical-incident Chat Memory, Project CodeGraph, Release-Checklist Skill\n```\n\nThe pitch: you’re not opening four disconnected chat windows, you’re assembling a squad that inherits the team’s accumulated experience — and a small team’s experience can keep compounding rather than resetting with every new session.\n\n## Governance: Private by Default, Sharing Is Explicit [#](#governance-private-by-default-sharing-is-explicit)\n\n| Visibility | Semantics |\n|---|---|\n`private` | Only the Owner can read — not even team admins |\n`team` | Team members can read; Owner/Admin can manage |\n`restricted` | Precise access via User / Role / Agent ACL |\n`agent` | Targeted equipping of specific agents on the same team |\n\nNew Chat Memory and Skills are private by default — sharing is an explicit action, not a default leak. This matters once a “memory hub” is holding real decisions and preferences: you can assign a Release Skill only to the Release Agent, an Architecture Wiki to all development agents, and CodeGraph specifically to Coder and Reviewer agents.\n\n## Technical Implementation [#](#technical-implementation)\n\n*Technical architecture overview — from github.com/TencentCloud/TencentDB-Agent-Memory*\n\nThe stated design goal isn’t “store everything” — it’s **what’s worth keeping, who can use it, and how to retrieve less while retrieving the right thing.**\n\n### Layered memory, not flat records [#](#layered-memory-not-flat-records)\n\n| Layer | What it stores | Primary use |\n|---|---|---|\n| L0 Conversation | Raw conversations, full context | Verify exact wording, timestamps, sources |\n| L1 Atom | Extracted facts, preferences, constraints, events | Precise recall of actionable information |\n| L2 Scenario | Knowledge blocks organized by project/scenario | Quickly restore a working context |\n| L3 Core / Persona | Long-term profiles, stable patterns | Rapid entry into a user’s/team’s context |\n\nRetrieval is layered too: L2/L3 provide a fast context bootstrap by default; when specific facts are needed, it falls back to BM25 + vector retrieval + RRF (Reciprocal Rank Fusion) across L1/L0, with results capped by item count, character budget, and timeout to keep memory from overwhelming the context window.\n\n### Memory as loadout, not global prompt injection [#](#memory-as-loadout-not-global-prompt-injection)\n\nChat Memory, Skill, Wiki, and CodeGraph are all registered uniformly as **Memory Assets**, and access is resolved via **Fixed Binding + ACL** — narrowing by Team, User, Agent, and visibility first, then retrieving based on the current query. Switching an agent or framework means re-equipping assets, not retraining.\n\n### Tool-based access, not wholesale injection [#](#tool-based-access-not-wholesale-injection)\n\nAgents discover capabilities via `/v3/tools/list`\n\n, then call `/v3/tools/call`\n\nto read specific Wiki pages, source code, or impact paths — documents and code are part of memory, but they stay as on-demand tools rather than being dumped into context wholesale.\n\n## Benchmark [#](#benchmark)\n\n| Benchmark | Without | With | Relative improvement |\n|---|---|---|---|\nPersonaMem | 48% | 76% | +59% |\n\nPersonaMem tests whether an agent correctly understands and applies user information after extended interactions. **This is a single benchmark reported by the maintainers**, not an independently reproduced result — a useful signal, not a guarantee it generalizes to your own workload.\n\n## Limitations (From the Project’s Own Notes) [#](#limitations-from-the-projects-own-notes)\n\n**Async processing delay**— Wiki and CodeGraph build asynchronously; allow time before they reach`ready`\n\nstatus**CodeGraph is public-repo-first**— private repositories and SSH credentials are “still being refined,” not fully supported yet** Manual asset binding**— the Hub supports manual binding today; fully automated memory routing is still under iteration** Limited framework support today**— OpenClaw, Hermes Agent, and SDK integration are supported now; broader cross-framework migration is on the roadmap, not shipped\n\n## Use Cases [#](#use-cases)\n\n### 1. Onboarding a New Agent (or Teammate) Without Re-Explaining Everything [#](#1-onboarding-a-new-agent-or-teammate-without-re-explaining-everything)\n\nImport existing docs, codebase, and past agent conversation sessions once — new team members and new agents both start from the “save file” instead of relearning the project from scratch.\n\n### 2. Role-Scoped Agent Teams [#](#2-role-scoped-agent-teams)\n\nGive a Reviewer agent CodeGraph and historical-incident Chat Memory, but not the Scout’s market-research Wiki — the loadout model keeps each agent’s context relevant instead of dumping everything into every agent.\n\n### 3. Governed Knowledge Sharing Across a Team [#](#3-governed-knowledge-sharing-across-a-team)\n\n`private`\n\n/`team`\n\n/`restricted`\n\n/`agent`\n\nvisibility lets an individual’s working notes stay private by default while explicitly promoting genuinely reusable Skills and Wiki pages to the team.\n\n### 4. Pre-Change Impact Analysis [#](#4-pre-change-impact-analysis)\n\nCodeGraph’s call-relationship and impact-path indexing lets an agent check what else might break before modifying shared code — closer to what a careful human reviewer would do than a plain text-match RAG lookup.\n\n## Related Repositories [#](#related-repositories)\n\n| Repository | Purpose |\n|---|---|\n|\n\n[CodeGraph (colbymchenry)](https://github.com/colbymchenry/codegraph)## Related Articles [#](#related-articles)\n\n[AI Agent Memory: Letta vs Mem0 vs A-Mem](https://dibi8.com/resources/llm-frameworks/ai-agent-memory-persistence-letta-mem0-a-mem-2026/)— for comparing single-agent memory frameworks against this team-level hub approach[Hermes Agent: Self-Improving AI Agent](https://dibi8.com/resources/llm-frameworks/hermes-agent-self-improving-ai-agent/)— one of the frameworks TencentDB Agent Memory integrates with directly\n\n## Conclusion [#](#conclusion)\n\n**TencentDB Agent Memory** targets a problem most agent-memory tools don’t: not “how does one agent remember one user,” but “how does a *team* of agents share governed, versioned experience without leaking everything to everyone.” The L0-L3 layered distillation, the four distinct asset types (especially Skill and CodeGraph, which go beyond what chat-log RAG models), and the explicit-sharing-by-default governance make it a more structured answer than most single-agent memory libraries — at the cost of currently narrower framework support (OpenClaw, Hermes, SDK) and features still labeled beta or roadmap.\n\n**Best for**: Teams running multiple agents (or agent + human teams) who need governed, shareable memory — not solo users who just want one agent to remember one conversation history.\n\n**GitHub**: [https://github.com/TencentCloud/TencentDB-Agent-Memory](https://github.com/TencentCloud/TencentDB-Agent-Memory)\n\n*Last updated: 2026-08-02*", "url": "https://wpnews.pro/news/tencentdb-agent-memory-a-team-level-memory-hub-not-just-per-agent-recall", "canonical_source": "https://dibi8.com/resources/llm-frameworks/tencentdb-agent-memory-team-memory-hub-2026/", "published_at": "2026-08-02 13:45:00+00:00", "updated_at": "2026-08-02 15:05:45.226531+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-agents", "ai-tools", "ai-infrastructure", "ai-research"], "entities": ["Tencent Cloud", "TencentDB Agent Memory", "GitHub", "Andrej Karpathy"], "alternates": {"html": "https://wpnews.pro/news/tencentdb-agent-memory-a-team-level-memory-hub-not-just-per-agent-recall", "markdown": "https://wpnews.pro/news/tencentdb-agent-memory-a-team-level-memory-hub-not-just-per-agent-recall.md", "text": "https://wpnews.pro/news/tencentdb-agent-memory-a-team-level-memory-hub-not-just-per-agent-recall.txt", "jsonld": "https://wpnews.pro/news/tencentdb-agent-memory-a-team-level-memory-hub-not-just-per-agent-recall.jsonld"}}