Examples

Integration Examples

Real-world use cases for Pulsar MCP Server in AI agents.

---

Trading Signals Agent

Monitor crypto communities for trading signals like token launches, airdrops, and partnerships.

Use Case

Your agent needs to:

• Monitor dozens of crypto channels 24/7

• Detect high-value trading signals

• Alert users in real-time

• Provide context and source links

MCP Tools Used

1. search_summaries_by_topic - Find mentions of keywords

2. semantic_search_messages - Get detailed context

3. get_channel_summary - Understand channel activity

Agent Instructions

You are a crypto trading signal monitor.

Every 30 minutes, search for:
- "token launch"
- "airdrop"  
- "partnership"
- "listing"

When you find matches:
1. Use semantic search to get message context
2. Check if signal is high-value (20+ mentions or high relevance)
3. Alert user with formatted signal:

🚨 SIGNAL: [Headline]
Channel: [Name]
Keywords: [comma-separated]
Timestamp: [ISO format]
Context: [2-sentence summary]

Example Output

🚨 Trading Signal Detected

Channel: Virtuals Protocol
Headline: MetaInsideAI Token Launch Announcement
Keywords: token, launch, airdrop, AI agent
Timestamp: 2025-01-15T14:00:00Z

Context: New AI agent token launching next week with airdrop to 
VIRTUAL stakers. Community showing strong bullish sentiment with 
127 new members joining in last 12 hours.

[View Channel] [Set Alert]

---

Research Assistant Agent

Answer user questions about crypto projects using community intelligence.

Use Case

Your agent needs to:

• Answer "What's happening with [project]?"

• Provide recent developments and sentiment

• Cite community sources

• Track multiple channels

MCP Tools Used

1. get_channel_summary - Get latest 12-hour summaries

2. semantic_search_messages - Find relevant discussions

3. list_all_channels - Discover available channels

Agent Instructions

You are a crypto research assistant.

When user asks about a project:
1. Get latest 3 summaries using get_channel_summary
2. Search for specific details using semantic_search_messages
3. Synthesize comprehensive answer

Format responses as:
📊 Key Developments (Last 36h):
[Numbered list with headlines and insights]

📈 Community Growth: [member delta]
💬 Sentiment: [bullish/bearish/mixed with reason]
🔑 Keywords: [top 5 keywords]

Source: [Channel name]
Last updated: [ISO timestamp]

Example Output

Based on the latest community activity (last 36 hours):

📊 Key Developments:

1. **MetaInsideAI 20x Growth**
   AI agent generated 20x returns in first week.
   Strong bullish sentiment in community.

2. **veVIRTUALS Staking Rewards**  
   Increased rewards for governance participants.
   APY details coming in next announcement.

3. **Gaming Partnership Talks**
   Unconfirmed reports of talks with major platform.
   Community speculating about integration timeline.

📈 Community Growth: +127 members (4,287 → 4,414)

💬 Sentiment: Highly bullish - excitement around agent 
   launches and staking opportunities

🔑 Keywords: AI, agent, staking, gaming, partnership

Source: Virtuals Protocol official channel
Last updated: 2025-01-15T18:00:00Z

---

Daily Briefing Bot

Generate daily intelligence briefings for subscribers.

Use Case

Your agent needs to:

• Compile daily crypto news

• Highlight active channels

• Track trending topics

• Deliver at scheduled time

MCP Tools Used

1. list_available_summaries - Get overview of all channels

2. get_channel_summary - Get details for active channels

3. search_summaries_by_topic - Track trending topics

Agent Instructions

You are a daily crypto briefing generator.

Every day at 7:00 AM UTC:
1. Use list_available_summaries to find channels with activity
2. Get summaries for top 10 most active channels
3. Group by category (AI, L1, DeFi)
4. Format as daily briefing

Template:
☀️ Crypto Community Briefing - [Date]

📰 Top Stories Today:
[For each channel: emoji icon, headline, sentiment, member growth]

🔥 Trending Topics:
[Top 5 keywords across all channels]

📊 Overall Sentiment: [analysis]

[Customize Channels] [View Full Details]

Example Output

☀️ Crypto Community Briefing - January 15, 2025

📰 Top Stories Today:

1. FETCH.AI 🤖
   → AWS Partnership Announced
   Community: 🔥 Very Bullish (+57 members)

2. VIRTUALS 🎮  
   → MetaInsideAI 20x Returns
   Community: 🚀 Extremely Active (+127 members)

3. SOLANA ⚡
   → Network Upgrade Scheduled
   Community: ⚠️ Mixed Sentiment (+23 members)

4. TONCOIN 💎
   → 15% Staking APY Launch
   Community: ✅ Positive Response (+45 members)

🔥 Trending Topics:
AI agents, staking, partnership, upgrade, rewards

📊 Overall Sentiment: Bullish across AI/agent sector, 
   mixed on infrastructure updates

[Customize Channels] [View Full Details]

---

Multi-Channel Topic Tracker

Track specific topics across all monitored communities.

Use Case

Your agent needs to:

• Monitor "staking rewards" across all channels

• Compare activity levels

• Identify trends

• Generate periodic reports

MCP Tools Used

1. search_by_timeframe - Search across time range

2. search_summaries_by_topic - Find keyword matches

3. semantic_search_messages - Get relevant context

Agent Instructions

You are a topic tracking agent.

When user requests topic tracking:
1. Use search_by_timeframe with user's topic and date range
2. Group results by channel
3. Count mentions and analyze sentiment
4. Generate comparative report

Format:
📊 Topic Report: "[topic]"
Period: [timeframe]

🔥 Most Active Channels:
[Ranked list with mention counts and key themes]

📈 Sentiment Trend: [direction and reasoning]

🔑 Key Themes:
[Bullet list of patterns]

💡 Insight: [1-sentence takeaway]

Example Output

📊 Topic Report: "staking rewards"
Period: Last 7 days

🔥 Most Active Channels:
1. Toncoin (23 mentions) - New 15% APY program
2. Solana (18 mentions) - Liquid staking updates  
3. Fetch.ai (12 mentions) - AI agent staking pool

📈 Sentiment Trend: ↗️ Increasingly positive

🔑 Key Themes:
• High APY competition (12-15% range)
• Liquid staking gaining popularity
• Governance participation incentives

💡 Insight: Multi-chain staking competition heating up.
         Users comparing yields across ecosystems.

---

Sentiment Tracker

Monitor sentiment changes for specific tokens or projects over time.

Use Case

Your agent needs to:

• Track sentiment for Virtuals Protocol

• Monitor member growth trends

• Detect narrative shifts

• Alert on major changes

MCP Tools Used

1. get_channel_summary - Get historical summaries

2. semantic_search_messages - Find sentiment indicators

Agent Instructions

You are a sentiment tracking agent.

For each channel you monitor:
1. Get last 10 summaries (5 days of 12-hour periods)
2. Extract member growth deltas
3. Track keyword frequency changes
4. Detect sentiment shifts

Alert when:
- Member growth >50 in single period
- Negative keywords spike (exploit, hack, outage)
- Sentiment reverses (bullish → bearish or vice versa)

Format alerts:
🚨 SENTIMENT ALERT: [Channel]
Change: [from] → [to]
Trigger: [reason]
Member Impact: [delta]
Action: [recommendation]

---

Best Practices

Combine Tools for Context

Don't rely on single tool - build complete picture:

1. Overview - list_available_summaries

2. Details - get_channel_summary

3. Deep dive - semantic_search_messages

4. Cross-reference - search_summaries_by_topic

Refresh Strategically

• Summaries: Generated every 12 hours (06:00 & 18:00 UTC)

• Check ~10 minutes after generation for fresh data

• Don't poll more than every 30 minutes

Handle Rate Limits

With bearer token you get 60 req/min:

• Batch queries when possible

• Stagger automated checks

• Cache results appropriately

Parse MCP Responses

MCP tools return formatted strings. Your agent should:

• Extract key information (channel names, timestamps, keywords)

• Structure for your application's needs

• Store for trend analysis if building historical views

---

Framework Integration

LangChain

Add Pulsar as tool to your LangChain agent:

from langchain.agents import initialize_agent
from langchain.agents import AgentType

# Your MCP client will handle tool discovery
# Just configure Pulsar in MCP settings

agent = initialize_agent(
    tools=[],  # MCP tools auto-discovered
    llm=your_llm,
    agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION
)

CrewAI

Add Pulsar to your crew's capabilities:

from crewai import Agent, Crew

crypto_analyst = Agent(
    role="Crypto Community Analyst",
    goal="Monitor communities for signals",
    # MCP tools available automatically
    backstory="Expert at analyzing community sentiment"
)

crew = Crew(agents=[crypto_analyst])

Eliza (ElizaOS)

Configure in your Eliza character file:

{
  "name": "CryptoAnalyst",
  "clients": ["discord", "twitter"],
  "mcp": {
    "servers": {
      "pulsar": {
        "type": "sse",
        "url": "https://mcp.askpulsar.com/mcp/sse"
      }
    }
  }
}

---

Testing Your Integration

1. Verify Connection

Ask your agent: "List all channels you can monitor"

Expected: Should list 25+ channels with names

2. Test Summary Retrieval

Ask: "Summarize Virtuals Protocol"

Expected: Should return formatted summary with developments

3. Test Search

Ask: "What are people saying about AI agents?"

Expected: Should return relevant messages with context

4. Verify Tools

Ask: "What Pulsar tools do you have access to?"

Expected: Should list all 7 tools with descriptions

---

Need Help?

📧 Email: [email protected] 💬 Telegram: @olenovyk 🐦 Twitter: @olenovyk

Share your implementation and we'll feature it here!