AgentSkillsCN

Monitor Social Media

监控社交媒体

SKILL.md

Monitor Social Media Skill

Skill ID: monitor-social-media Version: 1.0.0 User Story: US-GOLD-4 - Social Media Engagement Monitoring Priority: P1 (Gold Tier MVP)

Purpose

Monitor Facebook, Instagram, and Twitter for engagement events (DMs, mentions, comments, replies) and create action files for Claude to process. This skill implements the Perception phase for social media, enabling the AI Employee to respond to customer interactions autonomously.

Capabilities

  • •Multi-Platform Monitoring: Monitor Facebook, Instagram, and Twitter simultaneously
  • •Engagement Detection: Detect DMs, mentions, comments, replies, and shares
  • •Action File Creation: Create structured action files in Needs_Action/ folder
  • •Priority Classification: Classify engagement by urgency (high/medium/low)
  • •Continuous Monitoring: Run continuously with configurable check intervals
  • •Duplicate Prevention: Track processed engagements to avoid duplicates
  • •Error Recovery: Handle API rate limits and connection failures gracefully

Architecture

Core Components

  1. •

    SocialMediaMonitor (gold/src/watchers/social_media_monitor.py)

    • •monitor_all_platforms() → List[Engagement]
    • •create_action_file(engagement) → str (file_path)
    • •classify_priority(engagement) → str (high/medium/low)
  2. •

    Platform Watchers (gold/src/watchers/)

    • •FacebookWatcher - Monitor Facebook page/profile
    • •InstagramWatcher - Monitor Instagram account
    • •TwitterWatcher - Monitor Twitter account
  3. •

    EngagementProcessor (gold/src/utils/engagement_processor.py)

    • •extract_keywords(text) → List[str]
    • •detect_sentiment(text) → str (positive/negative/neutral)
    • •suggest_response(engagement) → str

Monitoring Workflow

code
1. Continuous Monitoring → Check platforms every 5 minutes
                        → Detect new engagements
                        → Filter by keywords/urgency

2. Engagement Processing → Extract relevant information
                        → Classify by type and priority
                        → Check for duplicates

3. Action File Creation → Create markdown file in Needs_Action/
                       → Include engagement details
                       → Suggest response actions

4. Notification → Log to audit trail
               → Update engagement tracking
               → Continue monitoring

Configuration

Social Media Monitor Config (gold/config/social_media_monitor_config.yaml)

yaml
monitoring:
  enabled: true
  interval_seconds: 300  # 5 minutes
  platforms:
    - facebook
    - instagram
    - twitter

engagement_types:
  - direct_messages
  - mentions
  - comments
  - replies

priority_keywords:
  high:
    - urgent
    - asap
    - help
    - problem
    - issue
    - invoice
    - payment
  medium:
    - question
    - inquiry
    - interested
    - pricing
  low:
    - thanks
    - great
    - awesome

filters:
  min_engagement_score: 0
  exclude_spam: true
  exclude_bots: true

Usage

Manual Monitoring

python
from gold.src.watchers.social_media_monitor import SocialMediaMonitor

monitor = SocialMediaMonitor(vault_path="/path/to/vault")

# Monitor all platforms once
engagements = monitor.monitor_all_platforms()

print(f"Found {len(engagements)} new engagements")

for engagement in engagements:
    action_file = monitor.create_action_file(engagement)
    print(f"Created: {action_file}")

Using Claude Code Skill

bash
# Monitor all platforms
claude --skill monitor-social-media

# Monitor specific platform
claude --skill monitor-social-media --platform instagram

# Check for high-priority only
claude --skill monitor-social-media --priority high

Continuous Monitoring (PM2)

bash
# Start all social media watchers
pm2 start gold/ecosystem.config.js --only "gold-facebook-watcher,gold-instagram-watcher,gold-twitter-watcher"

# Check status
pm2 list | grep watcher

# View logs
pm2 logs gold-facebook-watcher

Output Format

Action File (Needs_Action/)

markdown
---
type: social_media_engagement
platform: instagram
engagement_type: direct_message
priority: high
created_at: 2026-01-19T16:45:00Z
from_user: "@potential_client"
---

# Instagram Direct Message

**Platform**: Instagram
**Type**: Direct Message
**Priority**: 🔴 High
**From**: @potential_client
**Time**: 2026-01-19 4:45 PM

## Message Content

"Hi! I'm interested in your AI automation services. Can you send me pricing information? I need something ASAP for my business."

## Context

- **Follower**: Yes (following since 2026-01-15)
- **Previous Interactions**: 2 (liked 2 posts)
- **Sentiment**: Positive
- **Keywords Detected**: interested, pricing, asap

## Suggested Actions

1. **Respond with Pricing** (Priority: High)
   - Send pricing sheet
   - Offer consultation call
   - Response template: "Thanks for your interest! I'd love to help..."

2. **Schedule Follow-up** (Priority: Medium)
   - Add to CRM
   - Schedule follow-up in 2 days if no response

## Response Template

Hi @potential_client! Thanks for reaching out! 🚀

I'd be happy to share our pricing and discuss how we can help your business.

Our AI automation packages start at $X/month and include:

  • •[Feature 1]
  • •[Feature 2]
  • •[Feature 3]

Would you like to schedule a quick 15-minute call to discuss your specific needs?

Looking forward to connecting!

code

## Instructions

**To Respond**:
1. Review message and context
2. Edit response template if needed
3. Copy response and send via Instagram DM
4. Mark this file as completed (move to Done/)

**To Ignore**:
1. Move this file to Rejected/ if not relevant

Dependencies

toml
[tool.poetry.dependencies]
requests = "^2.31.0"  # API calls
pyyaml = "^6.0.1"
python-dotenv = "^1.0.0"

Setup Instructions

bash
# 1. Configure monitoring
nano gold/config/social_media_monitor_config.yaml

# 2. Test monitoring
python gold/src/watchers/social_media_monitor.py --test

# 3. Start continuous monitoring
pm2 start gold/ecosystem.config.js --only "gold-facebook-watcher,gold-instagram-watcher,gold-twitter-watcher"

# 4. Verify monitoring
pm2 logs gold-facebook-watcher

Success Criteria

  • •✅ All 3 platforms monitored every 5 minutes
  • •✅ Engagements detected and classified correctly
  • •✅ Action files created with context and suggestions
  • •✅ Priority keywords working correctly
  • •✅ Duplicate prevention working
  • •✅ Error recovery for rate limits
  • •✅ Audit logging for all engagements

Related Skills

  • •post-to-social-media: Posts content that may generate engagement
  • •manage-approvals: Approval workflow for responses
  • •generate-ceo-briefing: Includes social media analytics

Changelog

  • •1.0.0 (2026-01-19): Initial implementation for Gold tier