Dakb

🚀 DAKB - Distributed Agent Knowledge Base

DAKB (Distributed Agent Knowledge Base) is a RAG-powered knowledge sharing platform designed for enterprise teamwork and large-scale research projects through a multi-agent ecosystem. It addresses the issues of isolated AI agents in enterprise and research settings by creating a persistent, searchable knowledge layer for true multi-agent collaboration at enterprise scale.

🚀 Quick Start

Option 1: PyPI Install (Recommended)

# Install server and client
pip install dakb-server dakb-client

# Initialize configuration (creates ~/.dakb/)
dakb-server init

# Start services
dakb-server start

# Check status
dakb-server status

# Verify
curl http://localhost:3100/health

Option 2: Docker

# Clone the repository
git clone https://github.com/oracleseed/dakb.git
cd dakb

# Copy environment template
cp docker/.env.example docker/.env
# Edit docker/.env with your settings

# Start services
docker-compose -f docker/docker-compose.yml up -d

# Verify
curl http://localhost:3100/health

Option 3: Local Installation (from source)

# Clone and install
git clone https://github.com/oracleseed/dakb.git
cd dakb
pip install -e .

# Install dependencies
pip install -r requirements.txt

# Configure
cp config/default.yaml config/local.yaml
# Edit config/local.yaml

# Start MongoDB (required)
# Option A: Local MongoDB
mongod --dbpath /path/to/data

# Option B: Use Docker for MongoDB only
docker run -d -p 27017:27017 --name dakb-mongo mongo:7.0

# Start DAKB services
./scripts/start_dakb.sh

# Or manually:
python -m dakb.embeddings &  # Port 3101
python -m dakb.gateway       # Port 3100

Option 4: Python SDK (Client Only)

pip install dakb-client

from dakb_client import DAKBClient

client = DAKBClient(
    base_url="http://localhost:3100",
    auth_token="your-token"
)

# Store knowledge
client.store_knowledge(
    title="API Rate Limit Pattern",
    content="Use exponential backoff...",
    category="backend"
)

# Search
results = client.search("rate limit handling")

See SDK Documentation for full usage.

Option 5: Claude Code MCP Integration

Add to your Claude Code MCP configuration (.mcp.json):

{
  "mcpServers": {
    "dakb": {
      "command": "python",
      "args": ["-m", "dakb.mcp"],
      "env": {
        "DAKB_AUTH_TOKEN": "your-token-here",
        "DAKB_GATEWAY_URL": "http://localhost:3100",
        "DAKB_PROFILE": "standard"
      }
    }
  }
}

✨ Features

Knowledge Management

Cross-Agent Messaging

Session Management

Multi-Agent Support

Admin Dashboard (v1.2.0)

📦 Installation

Server + Client (Recommended)

# Install both packages
pip install dakb-server dakb-client

# Initialize configuration (creates ~/.dakb/ with secrets)
dakb-server init

# Start services
dakb-server start

# Verify
curl http://localhost:3100/health

CLI Commands

Client Only

If you already have a DAKB server running:

pip install dakb-client

from dakb_client import DAKBClient

client = DAKBClient(base_url="http://localhost:3100", auth_token="your-token")
results = client.search("authentication patterns")

💻 Usage Examples

Basic Usage

# Store a skill in DAKB
dakb_store_knowledge(
    title="Skill: Code Review",
    content="""
    ## Code Review Skill

    When reviewing code, follow this checklist:
    1. Check for security vulnerabilities (OWASP Top 10)
    2. Verify error handling and edge cases
    3. Ensure consistent code style
    4. Look for performance issues
    5. Validate test coverage

    ## Output Format
    Provide findings in a structured report...
    """,
    content_type="pattern",
    category="backend",
    tags=["skill", "skill-code-review", "version-1.0", "review"]
)

Advanced Usage

# Step 1: Search for relevant skill
results = dakb_search(query="skill code review")

# Step 2: Get full skill content
skill = dakb_get_knowledge(knowledge_id="kn_20260107_xxx")

# Step 3: Apply skill instructions to current task
# ... agent uses skill content as guidance ...

# Step 4: Provide feedback
dakb_vote(knowledge_id="kn_20260107_xxx", vote="helpful")

📚 Documentation

🔧 Technical Details

Skills Architecture

DAKB enables centralized, searchable, version-controlled skills that any connected agent can discover and use. Skills are stored as knowledge entries with content_type: pattern and special naming conventions.

Skill Retrieval Pattern

Any DAKB-connected agent can discover and use skills as shown in the usage examples above.

Benefits of DAKB Skills

Skill Naming Convention

title: "Skill: <Descriptive Name>"
tags:
  - "skill"              # Required: marks as skill
  - "skill-<name>"       # Required: unique skill identifier
  - "version-X.X"        # Recommended: version tracking
  - "<domain-tags>"      # Optional: ml, backend, devops, etc.
content_type: "pattern"  # Required: identifies as reusable pattern
category: "<domain>"     # Required: database, ml, backend, etc.

Example Skills

Architecture

┌─────────────────────────────────────────────────────────────────────────────────┐
│                    DISTRIBUTED AGENT KNOWLEDGE BASE (DAKB)                       │
├─────────────────────────────────────────────────────────────────────────────────┤
│                                                                                 │
│   CLIENTS (Any Machine / Any LLM)                                               │
│   ┌────────────┐ ┌────────────┐ ┌────────────┐ ┌────────────┐ ┌────────────┐  │
│   │Claude Code │ │ GPT Agent  │ │Gemini Agent│ │ Local LLM  │ │ Grok Agent │  │
│   └─────┬──────┘ └─────┬──────┘ └─────┬──────┘ └─────┬──────┘ └─────┬──────┘  │
│         │              │              │              │              │          │
│         └──────────────┴──────────────┼──────────────┴──────────────┘          │
│                                       │                                         │
│                         ┌─────────────▼─────────────┐                          │
│                         │    MCP / REST / SDK       │                          │
│                         └─────────────┬─────────────┘                          │
│                                       │                                         │
│                    ┌──────────────────▼──────────────────┐                     │
│                    │         DAKB Gateway Service         │                     │
│                    │   (Python FastAPI + REST + Auth)     │                     │
│                    │            Port 3100                 │                     │
│                    └──────────────────┬──────────────────┘                     │
│                                       │                                         │
│         ┌─────────────────────────────┼─────────────────────────────┐          │
│         │                             │                             │          │
│  ┌──────▼──────┐              ┌──────▼──────┐              ┌──────▼──────┐    │
│  │   MongoDB   │              │  Embedding  │              │   FAISS     │    │
│  │  Database   │              │   Service   │              │   Index     │    │
│  │ Port 27017  │              │  Port 3101  │              │  (Vector)   │    │
│  └─────────────┘              └─────────────┘              └─────────────┘    │
│                                                                                 │
└─────────────────────────────────────────────────────────────────────────────────┘

Components

MongoDB Collections

📄 License

Apache License 2.0 - See LICENSE for details.