Copy, paste, change the words in {curly brackets}. Field names match Cornersight's webhook and CSV export.
1. Build me a webhook receiver (Claude Code)
Build a small web server that receives Cornersight webhooks.
Each request is a POST with JSON: { event, data, timestamp }.
data has: leadId, engagementType (Like | Comment | Author), name, jobTitle,
companyName, companyDomain, companyStaffRange, country, linkedinUrl, email (may be null),
isIcp, commentText (may be null), postUrl, postText, trackedProfile.
1. Check the X-Cornersight-Signature header. It is "sha256=" + a hex
HMAC-SHA256 of "{X-Cornersight-Timestamp header}.{raw request body}",
using my signing secret from the env var CORNERSIGHT_SECRET. Use the raw
body, not re-encoded JSON. Reject if it doesn't match.
2. Skip any leadId I've already saved.
3. Save each lead as a row in {SQLite / Postgres / a CSV file}.
4. If isIcp is true and engagementType is Comment, also post a message to {Slack webhook URL env var}.
Write tests with two sample payloads. Explain how to run it locally.
2. Personalised first line (Clay AI column or any AI tool)
You write the first line of a cold email.
Person: {name}, {jobTitle} at {companyName}.
They engaged with a LinkedIn post. Post text: "{postText}"
Their comment (may be empty): "{commentText}"
Write one sentence, under 25 words, about the topic of the post.
Do NOT say you saw their like or comment. Do NOT flatter. No exclamation marks.
Sound like a peer who has an opinion on the topic.
3. Is this lead a fit? (Clay AI column or Claude)
My ideal customer: {e.g. founders and heads of sales at B2B companies with 10 to 500 staff}.
Lead: {jobTitle} at {companyName}, {companyStaffRange} staff, {companyIndustry}, {country}.
Answer with exactly one word, YES or NO, then one short reason.
4. Weekly content brain digest (Claude / ChatGPT)
Upload the "Content Brain" sheet (from the n8n workflow 03) first.
This sheet holds LinkedIn posts from my space this week.
Columns: postUrl, postedAt, keywordOrCreator, author, authorTitle, postText.
1. Group the posts into themes. For each: a one-line name, how many posts, and 2 to 3 exact quotes.
2. Which themes are new this week compared with {last week's digest, pasted below}?
3. Give me 3 post ideas for {my company / my role}, each with a hook in my buyers' own words.
4. List 5 authors I should start tracking, and why.
Ignore promotional posts and job ads.
5. Comment mining (Claude / ChatGPT)
Here are comments my target buyers left on LinkedIn posts this week (from the commentText column).
Tag each one: question, objection, story or agreement.
Then list the 5 most common questions in plain words. Those are my next 5 posts.
6. Analyse my signal sources (Claude Code, on a CSV export)
Read leads.csv (a Cornersight export). For each value of trackedProfile (the creator or keyword),
count: total leads, ICP leads (isIcp = true), share of ICP, and comments vs likes.
Sort by ICP leads. Tell me which 3 sources to keep, which to drop, and why.
Output a short table and 3 bullet points.