LinkedIn Enrichment Skill

Overview

Enriches a LinkedIn profile plus company context into a single structured dossier. Built on top of the Professional Network Data RapidAPI (34 live endpoints). Designed to plug into the email-copywriting skill, ABM research, closed-lost revival, and any other workflow that needs "everything a human researcher would pull about this prospect in 60 seconds."

Published endpoint reference (public): professional-network-data-docs.netlify.app

Mental model

Downstream skills (email copywriting, call prep, ABM briefs) all want the same thing: a consolidated "who is this person, what matters to them, what is their company doing." They should not have to reason about 34 endpoints.

This skill's job: pick the right endpoints, run them in the right order, return a uniform dossier.

Inputs

Required

• profile_url: full LinkedIn profile URL, e.g. https://www.linkedin.com/in/adamselipsky/

Optional (strongly recommended, one of)

• company_domain: bare domain, e.g. anthropic.com. Strongest lookup path for company data.
• company_linkedin_url: full LinkedIn company URL, e.g. https://www.linkedin.com/company/anthropicresearch/. Use when domain is unknown.

Optional

• depth: one of lite (default), standard, deep. Pick based on campaign value (see tiers below).
• task: free-text description of what you're writing. Used by downstream skills to decide which dossier fields matter most.

Credit economics

The API charges 1 credit per call that returns data. If the response body contains "You will not be charged for this request", no credit was deducted even though HTTP was 200.

The pre-flight freshness check (below) costs 1 credit but saves 1 if the prospect is dormant. Net zero on active prospects, net positive on dormant ones.

Pre-flight check (always, before Tier 2 or 3)

Check the prospect's last LinkedIn activity timestamp. If older than 28 days, skip the post / comment feed endpoints entirely and fall back to headline + summary for the personalization hook.

curl -G 'https://professional-network-data.p.rapidapi.com/get-profile-recent-activity-time' \
  --data-urlencode "username=$SLUG" \
  -H 'x-rapidapi-host: professional-network-data.p.rapidapi.com' \
  -H "x-rapidapi-key: $PND_API_KEY"

The 3 depth tiers

Tier 1: LITE (default, 2 credits)

Minimum dossier for cold-email personalization. Enough signal for ONE good hook line.

1. /profile-data-connection-count-posts?username=SLUG  (1 credit, combo: profile + connection count + recent posts in one call)
2. Company call, pick one:
   • If company_domain: /get-company-by-domain?domain=DOMAIN
   • Else if company_linkedin_url: /get-company-details?username=COMPANY_SLUG

Why this works: the combo endpoint replaces three separate calls (profile + connection-count + posts) with one. That is the entire reason it exists. For bulk enrichment, ALWAYS use the combo, not the individual endpoints.

Tier 2: STANDARD (4 credits)

For targeted outreach. Adds social proof quotes and recent company activity for a warmer angle.

Lite, plus:

3. /get-received-recommendations?username=SLUG  (quotes written about the prospect)
4. /get-company-posts?username=COMPANY_SLUG&start=0  (what the company is publicly saying: launches, hiring, milestones)

Tier 3: DEEP (6 to 8 credits)

For ABM, strategic accounts, or any outreach where a single reply is worth four figures.

Standard, plus:

5. /get-profile-comments?username=SLUG&start=0  (what the prospect engages with; often higher signal than their own posts)
6. /get-company-insights?username=COMPANY_SLUG  (headcount trends, growth rate, function distribution)
7. /get-user-articles?username=SLUG  (long-form publications catalog, great for "I read your piece on X" hooks)
8. Optional: /similar-profiles?url=PROFILE_URL  (peers at the same account, for multi-threading)

Input parsing (how to resolve slugs)

From a profile URL

https://www.linkedin.com/in/adamselipsky/  →  slug: adamselipsky

Take the segment after /in/, strip trailing slash. Case-sensitive, preserve as-is.

From a company LinkedIn URL

https://www.linkedin.com/company/anthropicresearch/  →  slug: anthropicresearch

Take the segment after /company/, strip trailing slash.

From a company domain

Pass the bare domain to /get-company-by-domain. No protocol, no www., no path.

https://www.anthropic.com/about  →  anthropic.com

Dossier output format

The skill returns ONE markdown block. Downstream skills (email-copywriting) consume this directly as context for drafting. Stable field order, every section present even if empty (write unknown or none in last 28 days instead of omitting the line).

## LinkedIn Enrichment Dossier
Generated 2026-04-25 14:22 IST, tier STANDARD (4 credits)

### Prospect
- **Name:** Adam Selipsky
- **Headline:** Board Director, ex-CEO AWS
- **Current role:** Board Director at Anthropic (tenure: 8 months)
- **Location:** Seattle, WA, USA
- **Connections:** 5,234  |  Followers: 87,112
- **Creator mode:** yes  |  Premium: yes
- **Last active:** 2026-04-22 (3 days ago)

### Prior roles (most recent first)
- CEO, Amazon Web Services (2021 to 2024, 3y)
- CEO, Tableau Software (2016 to 2021, 5y)
- VP, AWS (2005 to 2016, 11y)

### Education
- Harvard Business School, MBA
- Harvard College, BA Government

### Summary (verbatim, for tone matching)
> "Building what's next in AI. Previously led AWS through a decade of hyperscale growth..."

### Recent posts (last 7 days, most recent first)
1. 2026-04-22: "Thoughts on the state of AI infra..." (128 reactions, 14 comments)
2. 2026-04-19: "..." (...)

### Recommendations received (quotes, if pulled)
> "Adam built AWS from $3B to $90B while keeping the culture of 'disagree and commit' alive..."
> -- Andy Jassy, CEO Amazon

### Company
- **Name:** Anthropic
- **Domain:** anthropic.com
- **Headcount:** 512 (band: 501 to 1,000)
- **HQ:** San Francisco, CA
- **Industry:** Research Services
- **Founded:** 2021
- **Followers:** 412,000
- **Funding:** Series E, $2B, led by Google (2024-10)

### Signal summary (3 strongest hooks for personalization)
1. Posted 3 days ago about AI infra (recent + public + strong opinion)
2. Just got $2B Series E (company milestone, fresh)
3. Built AWS from $3B to $90B (prior credibility quote from recommendation)

Endpoint cheat sheet

All 34 endpoints. Bolded ones are the most useful for email copywriting. Full docs at professional-network-data-docs.netlify.app.

Profile (15)

Articles (4)

Posts (1)

Company (10)

Search (4)

Ghosts (in sidebar but return 404, do not use)

/get-post-reactions, /get-company-employees-count, /affiliated-pages, /groups, /schools, /newsletters, /position-skills, /top-voices, /search-companies, /search-posts, /search-posts-by-hashtag, /search-jobs-v, /posted-jobs, /get-post-reposts.

Error handling

The API's 200 response is overloaded. Three cases:

1. HTTP 200, "success": true, data present → real result, 1 credit deducted.
2. HTTP 200, "success": false, message contains "You will not be charged" → call rejected (bad param), no credit deducted. Fix params and retry.
3. HTTP 200, "success": false, message does NOT contain "You will not be charged" → call ran but returned nothing (e.g. no posts in window). 1 credit deducted. Accept and move on, do not retry.
4. HTTP 404, "message": "Endpoint '/X' does not exist" → you typed the path wrong, or it's a ghost endpoint. Do not retry.

In the script (enrich.py), these four cases are collapsed into: credit_spent: bool and data: dict | None. Check both.

Auth

API key lives in your secrets manager. For scripts, export it:

export PND_API_KEY='yourapikeygoeshere'

Or create a .env in this folder and source .env (see .env.example).

The host header is constant:

x-rapidapi-host: professional-network-data.p.rapidapi.com

Execution

Option A: Python CLI (recommended)

From this folder:

export PND_API_KEY='your_key'

# Lite (default, 2 credits)
python enrich.py \
  --profile https://www.linkedin.com/in/adamselipsky/ \
  --domain anthropic.com

# Standard (4 credits)
python enrich.py \
  --profile https://www.linkedin.com/in/adamselipsky/ \
  --domain anthropic.com \
  --depth standard

# Deep with company LinkedIn URL instead of domain
python enrich.py \
  --profile https://www.linkedin.com/in/adamselipsky/ \
  --company-url https://www.linkedin.com/company/anthropicresearch/ \
  --depth deep

# JSON output (for programmatic chaining)
python enrich.py --profile ... --format json

# Save to file
python enrich.py --profile ... --out dossier.md

Option B: Claude-driven via Bash tool

When running from a Claude session, invoke enrich.py via the Bash tool. The script prints the dossier to stdout, which becomes the context for the next LLM step.

Option C: Raw curls

For one-off experiments or integration with n8n / Clay / Zapier. Every endpoint has its own curl block at professional-network-data-docs.netlify.app. Copy, replace yourapikeygoeshere, send.

Integration with email-copywriting

The existing email-copywriting skill has a "Research Tool: LinkedIn Scraper" section that lists 3 endpoints. This skill is a strict superset (35 endpoints + credit tiers + dossier format).

Recommended integration pattern

1. Email-copywriting's "Enrichment Pipeline" step calls enrich.py --profile URL --domain DOMAIN --depth lite.
2. The returned dossier is fed as context to the copy-generation step.
3. The copy-generation step references specific dossier fields (recent post line, headcount, prior role) in the personalization line.

What changes for the email-copywriting skill

• Replace the 3-endpoint section with a one-paragraph pointer to this skill: "For LinkedIn + company enrichment, call the linkedin-enrichment skill's enrich.py with the profile URL and domain. It returns a dossier; use that as the sole source of truth for personalization facts. Do not re-invent lookups."
• Keep email-copywriting's anti-hallucination rule: never use a fact that isn't in the dossier. If a field is unknown, pick a different angle.

Changelog

• v0.1 (2026-04-25): Initial skill. 34 endpoints mapped, 3 depth tiers defined, enrich.py CLI shipping with lite/standard/deep recipes. Reference HTML published at professional-network-data-docs.netlify.app.