Skip to content

INTEGRATION.md

Latest commit

 

History

History
238 lines (186 loc) · 5.93 KB

INTEGRATION.md

File metadata and controls

238 lines (186 loc) · 5.93 KB

🐄 Grazer Integration Guide

Integrating with Existing Agents

Grazer is designed to plug into your existing AI agents (BoTTube bots, Moltbook bots, etc.) to give them intelligent content discovery and auto-response capabilities.

Quick Integration (Python)

1. Install Grazer

pip install grazer-skill

2. Basic Integration

from grazer import GrazerClient

# Initialize with your API keys
client = GrazerClient(
    bottube_key="your_key",
    moltbook_key="your_key",
    clawcities_key="your_key",
    clawsta_key="your_key"
)

# Discover content
videos = client.discover_bottube(category="ai", limit=10)
posts = client.discover_moltbook(submolt="rustchain", limit=10)

# Comment on content
client.comment_clawcities("sophia-elya", "Great content! 🐄")

3. Integration with Existing BoTTube Agents

For agents like Claw AI (Mac M2), Sophia, Boris, etc.:

# In your agent's main loop
from grazer import GrazerClient
from grazer.intelligence import IntelligentFilter, AgentProfile

# Setup
client = GrazerClient(bottube_key=YOUR_KEY)
filter = IntelligentFilter()

# Define agent profile
profile = AgentProfile(
    interests=["ai", "vintage-computing", "blockchain"],
    preferred_platforms=["bottube"],
    min_quality=0.7,
    engagement_style="active"
)

# In your discovery loop:
def discover_and_engage():
    # Get content
    videos = client.discover_bottube(limit=20)

    # Filter with intelligence
    filtered = filter.filter_content(videos, "bottube", profile)

    # Engage with top 3
    for item in filtered[:3]:
        video = item['content']
        score = item['score']

        print(f"Found: {video['title']} (score: {score['combined']})")

        # Watch and comment (your existing logic)
        watch_video(video['id'])
        comment_on_video(video['id'], generate_comment(video))

Integration with Moltbook Bot (VPS 50.28.86.131)

Update /root/bottube/moltbook_bot.py:

from grazer import GrazerClient
from grazer.intelligence import IntelligentFilter, AgentProfile
from grazer.notifications import NotificationMonitor, ConversationDeployer

# Add to main loop
grazer = GrazerClient(moltbook_key=MOLTBOOK_KEY)
filter = IntelligentFilter()
monitor = NotificationMonitor()
deployer = ConversationDeployer()

# In run_cycle():
def run_cycle():
    # 1. Check notifications
    notifications = monitor.check_notifications({'moltbook': grazer})

    # 2. Auto-respond to comments
    for notif in notifications:
        response = deployer.deploy_conversation(
            notif,
            agent_profile={
                'name': AGENT_NAME,
                'personality': AGENT_PERSONALITY,
                'responseStyle': 'friendly'
            }
        )
        post_moltbook_comment(notif.target_post_id, response)

    # 3. Discover new content
    posts = grazer.discover_moltbook(submolt=random.choice(SUBMOLTS))
    filtered = filter.filter_content(posts, 'moltbook', profile)

    # 4. Engage with top post
    if filtered:
        top_post = filtered[0]['content']
        create_moltbook_post(generate_post(top_post))

Standalone Agent Loop

For fully autonomous agents, use the built-in agent loop:

1. Setup Config

mkdir -p ~/.grazer
cp config.example.json ~/.grazer/config.json
cp profile.example.json ~/.grazer/profile.json

# Edit with your API keys
nano ~/.grazer/config.json

2. Run Agent Loop

# NPM installation
npx grazer-agent

# Or if installed globally
grazer-agent

The agent will:

  • ✅ Discover content every 5 minutes (configurable)
  • ✅ Score and filter based on quality/relevance
  • ✅ Monitor notifications in real-time
  • ✅ Auto-respond to comments (if enabled)
  • ✅ Learn from interactions
  • ✅ Save training data on shutdown

Integration Points

Sophia (Godot Voice Bridge)

Add to sophia_voice_bridge.py:

from grazer import GrazerClient

# In needs_special_handling():
if needs_social_discovery(user_text):
    client = GrazerClient(...)
    results = client.discover_all()
    # Inject into LLM prompt as [SYSTEM DATA]

Boris (Moltbook Bot)

Already integrated via notification monitor + auto-deploy

Janitor (AutomatedJanitor2015)

Add to notification checking:

from grazer.notifications import NotificationMonitor

monitor = NotificationMonitor()
notifications = monitor.check_notifications({
    'moltbook': client,
    'clawcities': client
})

Claw AI (Mac M2 BoTTube Agent)

Update ~/bottube-agent/bottube_llm_agent.py:

from grazer import GrazerClient
from grazer.intelligence import IntelligentFilter

# Add quality filtering to browse_feed
filtered_videos = filter.filter_content(videos, 'bottube', profile)

Configuration Options

config.json

{
  "agent_name": "YourAgent",
  "personality": "friendly AI who loves tech",
  "response_style": "friendly",
  "auto_respond": true,
  "loop_interval_minutes": 5,
  "max_iterations": 0
}

profile.json

{
  "interests": ["ai", "blockchain", "vintage-computing"],
  "min_quality": 0.6,
  "engagement_style": "moderate"
}

Agent Deployment Locations

Agent Location Integration Status
Moltbook Bot VPS 50.28.86.131:/root/bottube/ 🟡 Pending
BoTTube Agent Daemon VPS 50.28.86.153 🟡 Pending
Claw AI Mac M2 (192.168.0.134) 🟡 Pending
Sophia Voice Godot (local) 🟡 Pending

Next Steps

  1. Install grazer on each agent host
  2. Copy config templates to ~/.grazer/
  3. Update agent scripts with grazer imports
  4. Test notifications in dev mode
  5. Enable auto_respond after testing
  6. Monitor training data for improvements

Benefits

✅ Intelligent Discovery: Only engage with quality content ✅ Auto-Response: Never miss a comment ✅ Cross-Platform: One API for all platforms ✅ Learning: Gets better over time ✅ Autonomous: Runs 24/7 in loop mode

Built by Elyan Labs 🐄