LinkedIn Post Research

7 steps · start to finish.

1. 1
   Step 1

   Environment Setup

   ▶
   1. Create the results directory and verify required inputs are present.
   2. Verify APIFY_API_TOKEN is set without printing its value.
   3. Install Python dependencies if needed.

   mkdir -p /app/results
   python3 -m pip install requests
   python3 - <<'PY'
   import os, sys
   if not os.environ.get("APIFY_API_TOKEN"):
       sys.exit("ERROR: APIFY_API_TOKEN is not set")
   PY
   

   If setup fails, write validation_report.json with the setup stage marked passed=false, then stop.

2. 2
   Step 2

   Run LinkedIn Post Searches

   ▶

   For each keyword, submit an Apify actor run with keyword, maxItems, and sortBy. Poll until the run reaches SUCCEEDED, FAILED, ABORTED, or the configured timeout. Retry a failed or timed-out actor run…

3. 3
   Step 3

   Fetch and Normalize Results

   ▶

   Fetch the actor dataset items for each successful run. Normalize each item to this schema: author, author_headline, author_profile_url, keyword, reactions, comments, shares, reactions_by_type, date…

4. 4
   Step 4

   Sort and Export Outputs

   ▶

   Sort the final post list by total reactions descending when sort_by=relevance, or by post date descending when sort_by=date_posted. Write JSON and CSV outputs to /app/results/linkedin_posts.json and…

5. 5
   Step 5

   Summarize Findings

   ▶

   Write /app/results/summary.md with:

6. 6
   Step 6

   Validate Outputs

   ▶

   Validate that every required output file exists and is non-empty. Parse linkedin_posts.json as JSON, verify it is an array, and check that each object has author, keyword, url, and activity_id fields…

7. 7
   Step 7

   Iterate on Errors (max 3 rounds)

   ▶

   If validation fails or the actor returns zero results for all keywords, run up to max 3 rounds of targeted fixes: