social-intel

name	social-intel
description	Use only when the user explicitly requests social-profile, competitor-account, engagement-pattern, content-opportunity, or cross-platform growth analysis and actionable coaching.

Social Media Intelligence

Use this playbook for requested social analysis and actionable growth coaching across one account, a competitor, or several platforms.

---

Tool

social_intel(platform, handle, mode?, competitor?)

• platform: instagram | tiktok | x | twitter | linkedin | facebook
• handle: with or without @
• mode: full (default — tries API first, falls back to scraping) | quick (scrape only)
• competitor: true to analyze a competitor without saving to entity files

---

Workflow

Own Account Analysis

1. social_intel({ platform: "instagram", handle: "@yourbrand" })
2. Report is returned to chat + saved to entities/social/instagram.md
3. Suggest 3 specific next actions based on the data

Competitor Analysis

1. social_intel({ platform: "instagram", handle: "@competitor", competitor: true })
2. Compare against own account data from entities/social/instagram.md
3. Identify gaps and opportunities

Full Social Audit (multiple platforms)

Run social_intel for each platform the user is on.
Synthesize into a cross-platform report covering:
- Which platform has best engagement rate
- Which content type performs best per platform
- Posting frequency comparison
- Growth trajectory per platform

---

Connecting Official APIs (for full analytics)

Without an API token, Prom uses public scraping (follower count, bio, post grid only). With an API token, Prom gets: every post's likes/comments/saves/reach/impressions/watch time.

Instagram

1. Go to developers.facebook.com → Create App → Instagram Graph API
2. Get a long-lived access token
3. In Prometheus: vault set social_instagram_token <token>

X (Twitter)

1. developer.twitter.com → Create project → Get Bearer Token
2. vault set social_x_token <bearer_token>

TikTok

1. developers.tiktok.com → Create app → Get access token
2. vault set social_tiktok_token <token>

LinkedIn

1. linkedin.com/developers → Create app → Marketing Developer Platform
2. vault set social_linkedin_token <token>

---

Analysis Framework

When delivering a social coaching report, always cover:

1. Audience Health

• Follower count + growth rate (if historical data available)
• Following ratio (high following/follower ratio = vanity follows)
• Engagement rate = (likes + comments) / followers × 100

2. Content Performance

• Top 3 posts and what made them work (timing, format, caption, topic)
• Bottom 3 posts — what to avoid
• Average engagement rate vs platform benchmarks:
  • Instagram: 1-3% = average, 3-6% = good, 6%+ = excellent
  • TikTok: 4-9% = average, 9%+ = good
  • X: 0.5-1% = average, 1%+ = good

3. Content Patterns

• Best performing content type (Reels vs Posts vs Stories / video vs image)
• Caption length correlation with performance
• Hashtag usage effectiveness
• Best posting times (if timestamp data available)

4. Growth Trajectory

• Is engagement rate stable, growing, or declining?
• Follower growth signals

5. Actionable Recommendations (always 5 specific actions)

• Numbered, prioritized by impact
• Each one must be specific and executable: not "post more" but "post 3x/week at 7pm EST based on your top 5 posts all being published in that window"

---

Entity File Format

Results are saved to workspace/entities/social/[platform].md. Read this file before responding to follow-up questions about the same platform. Update the "Last Updated" and "Followers" fields after each analysis run.

---

Output Format

Always structure the report as:

1. Profile snapshot (followers, following, posts, bio)
2. Performance summary (avg engagement, top posts, content breakdown)
3. What's working (2-3 specific observations)
4. What's not working (2-3 specific observations)
5. 5 priority actions (numbered, specific, executable)
6. Data source (API / scraped / partial)

Keep it direct. No padding. Numbers over adjectives.