🧠 Hermes + TencentDB

Give your AI assistant a long-term memory — no coding required.

In this guide, you'll connect Hermes Agent (an open-source AI assistant) to TencentDB Agent Memory (the memory system from our previous article). When connected, your AI assistant will:

Hermes Agent is a free, open-source AI assistant created by Nous Research. Think of it as an alternative to ChatGPT or Claude, but one that:

• Runs on your computer (your data stays private)
• Works with any AI model (OpenAI, Anthropic, Google, local models, and 20+ others)
• Has tools — it can read files, run commands, search the web, and more
• Is self-improving — it learns skills from experience

When you connect it to TencentDB, it gains a persistent memory that survives across sessions.

You → Hermes Agent → TencentDB Memory Core → AI Model (extracts memories) → Stored
                       ↑                                                      |
                       └──────────── Memories loaded next session ────────────┘

1. You talk to Hermes.
2. Hermes sends the conversation to TencentDB.
3. TencentDB uses an AI model to extract important facts.
4. Those facts are stored permanently.
5. Next time you talk to Hermes, it loads those memories automatically.

Step 1

Open your terminal and run the installer:

curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash

This installs Hermes and all its dependencies. After installation, verify it works:

hermes --version

You should see a version number.

Step 2

Hermes needs an AI model to think. Run the setup wizard:

hermes setup

This walks you through:

• Choosing an AI provider (OpenAI, Anthropic, Google, etc.)
• Entering your API key
• Selecting a model (e.g., gpt-4o, claude-sonnet-4-20250514)

You can also do this manually:

hermes model

Step 3

Hermes uses "plugins" to add new capabilities. The TencentDB memory plugin tells Hermes how to talk to your memory system.

First, locate the plugin files. If you followed the TencentDB setup guide, they're in your project folder:

# Copy the plugin into Hermes
cp -r ~/tencent-agent-memory/hermes-plugin/memory_tencentdb ~/.hermes/plugins/memory_tencentdb

Then enable it:

hermes plugins enable memory_tencentdb

Step 4

Hermes needs to know where to find your memory system. Add these settings to your Hermes configuration:

# Set the memory provider
hermes config set memory.provider memory_tencentdb

# Enable memory
hermes config set memory.memory_enabled true

Now Hermes knows to use TencentDB for memory, but it still needs to know where TencentDB is running.

Step 5

This is the crucial step. Hermes needs three pieces of information to connect:

1. Where TencentDB is running (the network address)
2. Which team the memories belong to
3. Which agent is doing the talking

5a. Find Your TencentDB Address

If TencentDB is running on the same computer as Hermes (the most common setup), the address is:

tdai-memory-core:8420

If TencentDB is on a different computer, use that computer's IP address instead (e.g., 192.168.1.100:8420).

5b. Create a Team and Agent in the Dashboard

Open your TencentDB dashboard at http://localhost:8125 and:

1. Go to Organization → Teams and click Create Team. Name it something like "My Projects".
2. Go to Agents and click Create Agent. Name it "Hermes Assistant".
3. Note down the IDs that appear — they look like team-abc123 and agt-xyz789.

5c. Configure Hermes

Add the connection details to your Hermes environment. Open ~/.hermes/.env and add:

MEMORY_TENCENTDB_GATEWAY_HOST=tdai-memory-core
MEMORY_TENCENTDB_GATEWAY_PORT=8420
MEMORY_TENCENTDB_TEAM_ID=team-abc123
MEMORY_TENCENTDB_AGENT_ID=agt-xyz789
MEMORY_TENCENTDB_USER_ID=usr-def456

Replace the IDs with the ones you noted from the dashboard.

Step 6

Restart Hermes to apply the changes:

# If running as a Docker container
docker compose restart

# If running directly
# Just start a new session

Now start a conversation with Hermes and tell it something about yourself:

Hermes will respond normally. But behind the scenes, it's also sending this conversation to TencentDB.

Step 7

After your conversation, open the TencentDB dashboard at http://localhost:8125:

1. Go to Chat Memory in the sidebar.
2. You should see a new entry called "Memory of Hermes Assistant".
3. Click on it — you'll see extracted facts from your conversation.

From now on, every conversation with Hermes will:

1. At the start: Hermes loads your existing memories (preferences, past projects, learned skills).
2. During the conversation: Hermes uses those memories to give you more personalized, context-aware responses.
3. At the end: The conversation is sent to TencentDB, which extracts new facts and updates your memory.

Over time, Hermes builds a rich understanding of:

One of TencentDB's most powerful features is team memory sharing. If you have multiple AI agents (or multiple people on your team), they can all share the same memory pool.

To set this up:

1. In the TencentDB dashboard, create a Team that multiple agents belong to.
2. When creating memories or skills, set their visibility to Team instead of Private.
3. All agents in that team automatically see shared memories.

This means:

• Agent A learns a useful procedure → Agent B can use it immediately.
• You tell one agent about a project → all agents know about it.
• Team conventions are enforced consistently across all agents.

# Check memory configuration
hermes config get memory

# List installed plugins
hermes plugins list

# View Hermes logs (look for memory-related messages)
docker compose logs gateway | grep -i memory

# Test the connection manually
curl http://localhost:8420/health

Congratulations! You now have:

• ✅ Hermes Agent — a powerful, private AI assistant
• ✅ TencentDB Agent Memory — a persistent memory system
• ✅ Connected together — your AI remembers everything that matters

Your AI assistant now has something most AI assistants lack: continuity. It grows with you, learns from you, and becomes more helpful over time.

• Explore the dashboard: Browse your extracted memories at http://localhost:8125
• Create skills: When Hermes does something useful, save it as a reusable skill
• Add more agents: Connect additional Hermes instances or other AI agents to the same memory pool
• Set up team sharing: Share memories and skills across your team

What will you teach it?