AgentDB
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@AgentDBawaken agent_1 with tool history"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
AgentDB
MCP-based agent memory management system with layered awakening.
Each agent gets its own SQLite database. An MCP server exposes tools for managing context (system prompts, todos, memories, skills, buffers, tool calls). A Flask dashboard provides a web UI and REST API for monitoring and administration.
Architecture
Worker Agent ──► MCP Server (agent_id scoped tools)
│
▼
SQLite per agent ◄── Dashboard (read-only views + admin API)
(WAL mode)Key design decisions:
Layered awakening — loads ~5K tokens instead of 50K+ by selecting top priorities, top memories, skill metadata only, and buffer titles only
Per-agent isolation — each agent has its own
.dbfile; no cross-agent data leakageWAL mode — concurrent readers never block writers
Auto-logging —
AutoLoggingMCPproxy records every tool call with timing, input/output summaries, without blocking tool execution
Related MCP server: yantrikdb-mcp
Project Structure
agent-db/
├── agentdb_server/ # MCP server
│ ├── server.py # Entry point, tool registration
│ ├── db.py # DatabaseManager (connections, migrations)
│ ├── schema.sql # SQLite schema (7 tables)
│ ├── tool_logger.py # AutoLoggingMCP proxy
│ └── tools/
│ ├── awaken.py # Layered awakening (6 layers)
│ ├── todo.py # Todo CRUD
│ ├── memory.py # Memory store & search
│ ├── skill.py # Skill catalog
│ ├── buffer.py # Long document storage
│ └── tool_call.py # Tool call logging & stats
├── dashboard/ # Flask web app
│ ├── app.py # Routes (index, agent detail, admin)
│ ├── api.py # REST API blueprint (/api)
│ ├── models.py # SQLAlchemy ORM models
│ └── templates/
│ ├── base.html # Dark theme base layout
│ ├── index.html # Agent list
│ ├── agent_detail.html # Agent overview
│ ├── awakening_detail.html
│ └── admin.html # Cross-agent admin overview
├── tests/
│ ├── test_core.py # DB layer + schema tests
│ └── test_api.py # REST API endpoint tests
└── pyproject.tomlQuick Start
Install
pip install -e ".[dashboard,dev]"Run MCP Server
# Default agents directory: ./agents
agentdb-server
# Custom directory
AGENTDB_AGENTS_DIR=/path/to/agents agentdb-serverRun Dashboard
agentdb-dashboard
# Serves at http://localhost:5000Run Tests
pytest tests/ -vMCP Tools
Awakening
Tool | Description |
| Load 6-layer context (~5K tokens) |
| Create new prompt version |
| Get active system prompt |
Awakening layers:
Layer | Content | Budget |
1 | Active system prompt | ~500 tokens |
2 | Top 5 pending todos by priority | ~200 tokens |
3 | Skills catalog (metadata only) | ~1000 tokens |
4 | Top 10 memories by importance | ~3000 tokens |
5 | Buffer references (titles only) | minimal |
6 | Recent 20 tool calls (optional) | ~300 tokens |
Todos
Tool | Description |
| Add todo (priority 1-10) |
| List todos, filter by status |
| Mark todo as done |
Memories
Tool | Description |
| Store memory (importance 0.0-1.0, type: fact/experience/insight) |
| Keyword search, ordered by importance |
| Get single memory |
Skills
Tool | Description |
| Add to catalog |
| Metadata by category (no full_doc) |
| Load full documentation |
Buffers
Tool | Description |
| Store long document |
| Load full content |
| List titles only |
Tool Calls
Tool | Description |
| Manually log external tool usage |
| Query call history |
| Per-tool stats: count, success/error rate, avg duration |
Dashboard REST API
Base URL: http://localhost:5000/api
Agents
GET /api/agents # List all agents with statsPer-Agent Resources
Each resource supports full CRUD at /api/agents/{id}/{resource}:
GET /api/agents/{id}/memories # List
GET /api/agents/{id}/memories/{mid} # Detail
POST /api/agents/{id}/memories # Create
PUT /api/agents/{id}/memories/{mid} # Update
DELETE /api/agents/{id}/memories/{mid} # DeleteSame pattern applies to: todos, skills, buffers, system_prompts.
Tool Calls & Awakenings (read-only)
GET /api/agents/{id}/tool_calls?tool_name=&limit= # Call history
GET /api/agents/{id}/tool_calls/stats # Aggregated stats
GET /api/agents/{id}/awakenings # Awakening history
GET /api/agents/{id}/awakenings/{aid} # Awakening detailExamples
# List agents
curl http://localhost:5000/api/agents
# Create a memory
curl -X POST http://localhost:5000/api/agents/my-agent/memories \
-H 'Content-Type: application/json' \
-d '{"title": "API pattern", "content": "Use REST conventions", "importance": 0.8}'
# Get tool call stats
curl http://localhost:5000/api/agents/my-agent/tool_calls/statsDatabase Schema
Each agent's SQLite database contains 7 tables:
Table | Purpose |
| Versioned behavioral guidelines |
| Task items with priority (1-10) and status |
| Knowledge notes with importance (0.0-1.0) and type |
| Skill catalog with metadata + full documentation |
| Long document storage |
| Tool invocation history with timing and status |
| Snapshots of what was loaded during each awakening |
Auto-Logging
The AutoLoggingMCP proxy wraps tool registration so that every tool call is automatically recorded:
Timing — duration in milliseconds
Input summary — JSON with large fields truncated to 200 chars
Output summary — truncated to 500 chars
Silent failure — logging errors never block tool execution
Exclusions —
log_tool_call,list_tool_calls,get_tool_statsare excluded to prevent recursion
Claude Code Integration
Add to your Claude Code MCP config:
{
"mcpServers": {
"agentdb": {
"command": "agentdb-server",
"env": {
"AGENTDB_AGENTS_DIR": "/path/to/agents"
}
}
}
}License
MIT
This server cannot be installed
Maintenance
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Related MCP Servers
- -license-quality-maintenanceA sophisticated MCP server providing advanced memory capabilities with RAG, hallucination detection, and enterprise-grade AI infrastructure for intelligent agent ecosystems.Last updated

yantrikdb-mcpofficial
Alicense-qualityAmaintenanceMCP server providing cognitive memory tools (remember, recall, think, etc.) for AI agents, enabling forgetting, consolidation, and contradiction detection.Last updated168AGPL 3.0- AlicenseAqualityDmaintenanceMCP server for long-term agent memory, providing persistent memory, searchable knowledge, and evolving identity for AI agents.Last updated53Apache 2.0
- Alicense-qualityCmaintenanceAn MCP server that provides persistent long-term memory for AI agents via local SQLite storage with low token overhead, enabling memory storage, retrieval, and management across sessions.Last updated1MIT
Related MCP Connectors
Cloud-hosted MCP server for durable AI memory
A paid remote MCP for agent memory MCP, built to return verdicts, receipts, usage logs, and audit-re
User-owned memory for AI agents, Copilot, Claude, IDEs, CLIs, and chat apps over remote MCP.
Latest Blog Posts
- Who's Calling? MCP Hosts Are an Identity Blind Spot (And the Spec Knows It)By Om-Shree-0709 on .mcpAgent IdentityOAuth 2.1
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/wcAmon/agent-db'
If you have feedback or need assistance with the MCP directory API, please join our Discord server