linkedin-research

linkedin-research — LinkedIn Profile Intelligence

---

Step 1: Parse & Setup

1. Extract the LinkedIn profile URL from $ARGUMENTS (strip whitespace).
2. Extract the slug from the URL: split on /in/, take the part after it, strip trailing / and query params. Example: https://www.linkedin.com/in/some-person/ → some-person.
3. Set paths:
   • PROSPECT_DIR = prospects/{slug}/
   • RAW_DATA = prospects/{slug}/linkedin_raw_data.json
   • OUTPUT = prospects/{slug}/notes/linkedin-intake.md
   • COOKIES = plugins/linkedin-research/.cookies.json
   • SCRAPER = plugins/linkedin-research/linkedin-research.py
4. Create directories: prospects/{slug}/notes/ (use mkdir -p).

IMPORTANT: Before running the scraper, check if linkedin_raw_data.json already exists at the output path. If it does, ask the user:

"linkedin_raw_data.json already exists for {slug}. Use existing data or re-scrape?" If they say use existing, skip Step 2 entirely.

---

Step 2: Run Scraper

Run the Playwright scraper via Bash:

cd /Users/ketankhairnar/Desktop/AIC && python plugins/linkedin-research/linkedin-research.py {profile_url} {RAW_DATA} --cookies {COOKIES}

• If exit code != 0, show the error and STOP. Do not proceed with missing data.
• If successful, read linkedin_raw_data.json and report:

✓ Scraper complete
  Profile: {meta.profile_text_length} chars
  Activity: {meta.activity_text_length} chars
  Articles: {meta.articles_text_length} chars
  Status: {scrape_status summary}

---

Step 3: Read Raw Data

Read the full linkedin_raw_data.json into context. You need ALL of it for the analysis passes.

DATA QUALITY GATE: Check meta.scrape_status and text lengths:

• If profile_text_length < 200 or scrape_status.profile is "failed" or "suspicious": STOP and tell the user "Scraper returned insufficient profile data — likely auth issue, private profile, or CAPTCHA. Re-export cookies: python plugins/linkedin-research/linkedin-research.py --export-cookies plugins/linkedin-research/.cookies.json"
• If activity_text_length < 100: Note "No activity data — this person may not post on LinkedIn. Analysis will rely on profile data only."
• If articles_text_length < 100: Note "No published articles found."