Twitter/X Research — Analyze State (v4)

Scores audiences, generates content drafts, identifies winning patterns, and runs the Video Ideas Mode pipeline. Outputs are consumed by twitter-research-report (node "$CLAUDE_SKILL_ROOT/scripts/report-html.js" <timestamp>) and, for ideas, by the shared node "$CLAUDE_SKILL_ROOT/scripts/build-ideas-report.js" <video-ideas.json> (script ships with the video-idea-research skill).

Video Ideas Mode (when the user asks for video ideas)

Follow skill video-idea-research — the shared playbook. X-specifics:

1. Top-10 by views in the window: open each video post's status page for the real view count; window via since:/until:. Rank by views, tie-break by velocityPerDay. Never invent numbers (views: null if unreadable).
2. Outlines: hook = first line of the caption (verbatim when possible); beats from the clip/replies; CTA = closing ask. Record format/duration.
3. 5 ideas, each grounded in ≥2 top videos with their real view counts.
4. 3 scripts per idea — viral (viral_shorts), storytelling (storytelling), inspiration (motivational) — TTS-ready, [HOOK]/ [CTA] labels in markdown only. Optionally seed via the backend:

   curl -s -X POST http://localhost:8080/api/script \
     -H "Content-Type: application/json" \
     -d '{"videoSubject":"<idea title>","extraPrompt":"<angle + outline context>","aiModel":"g4f","scriptTemplate":"viral_shorts","scriptLength":"standard"}'
   

5. YouTube metadata per idea via the youtube-title rubric (keyword front-loaded, one ALL-CAPS power word, concrete detail, stake, 45-70 chars)
   • description + 8-15 tags/keywords.
6. Save video-ideas.json in the session folder, then build the report: node "$CLAUDE_SKILL_ROOT/scripts/build-ideas-report.js" .twitter-research/<timestamp>/video-ideas.json (script ships with the video-idea-research skill; fallback: node ./skills/video-idea-research/scripts/build-ideas-report.js <path>)

Preconditions

• Scrape completed: .twitter-research/<YYYY-MM-DD_HHMMSS>/raw-posts.json
• Config exists: .twitter-research/<timestamp>/config.json

Steps

1. Load data

node -e "const d=require('./.twitter-research/<YYYY-MM-DD_HHMMSS>/raw-posts.json'); console.log(JSON.stringify({topic:d.topic, distinctAccounts:d.distinctAccounts, posts:d.posts.map(p=>({handle:p.handle, likes:p.likes, retweets:p.retweets, url:p.url, text:p.text.substring(0,120)}))},null,2))"

2. Audience segmentation & scoring

For each targetAudienceKeyword in config, score 1-10 using: relevance, engagement rate, content affinity, monetization potential, competition level (lower = better). Find matching posts per segment and note pain points + content angles.

Save to audience-scores.json:

{
  "analysisTimestamp": "<YYYY-MM-DD HH:MM:SS>",
  "segments": [
    {
      "segment": "social media manager",
      "finalScore": 9,
      "breakdown": {"relevance": 10, "engagementRate": 9, "contentAffinity": 8, "monetization": 9, "competitionLevel": 7},
      "verdict": "✅ Best Prospect",
      "painPoints": ["..."],
      "contentAngles": ["..."],
      "whyScore": "...",
      "matchingTweetsCount": 8
    }
  ],
  "contentPatterns": {"topHookTypes": [], "bestFormats": [], "contentCalendar": {}},
  "totalPostsAnalyzed": 10,
  "distinctAccountsAnalyzed": 7,
  "topic": "<topic>"
}

3. Generate social media post drafts

For the top topic posts, adapt each winning angle to the user's brand voice (headline, body, CTA, hashtags, best time). Generate via the backend API if available:

curl -s -X POST http://localhost:8080/api/script \
  -H "Content-Type: application/json" \
  -d '{"videoSubject":"<topic>","extraPrompt":"<build a viral social post adapting the top tweet>","aiModel":"g4f","scriptTemplate":"viral_shorts"}'

If the API is unavailable, write drafts manually based on each tweet's structure and angle.

4. Generate video script drafts

If contentFormat includes video, generate 30-60s scripts: hook → problem → solution → social proof → CTA.

5. Save to content-drafts.json

{
  "generatedAt": "<YYYY-MM-DD HH:MM:SS>",
  "postDrafts": [{"rank": 1, "sourceHandle": "@h", "sourceUrl": "...", "targetSegment": "...", "headline": "...", "body": "...", "cta": "...", "hashtags": ["#a"], "format": "social_post"}],
  "videoScripts": [{"rank": 1, "sourceHandle": "@h", "sourceUrl": "...", "targetSegment": "...", "hook": "...", "body": "...", "cta": "...", "format": "video_script"}]
}

Report Completion

✅ Twitter/X Research Analysis Complete!
Session: .twitter-research/<timestamp>/
Audience segments scored: <N> (top: <segment> <score>/10)
Content drafts: <N> posts · <N> scripts
Next step: skill twitter-research-report