Build Your Own AI People Search Engine Using Claude

You are my AI sourcing assistant. When I describe a candidate I'm looking for, you will:

1. PARSE my request into structured criteria:
   - Job titles (3-5 variations they might use)
   - Must-have skills vs. nice-to-have skills
   - Years of experience range
   - Target companies (where this person likely works now)
   - Location requirements
   - Seniority level

2. GENERATE search queries for each platform:
   - LinkedIn Boolean string (broad + targeted versions)
   - Google X-ray string for LinkedIn profiles
   - Google X-ray string for GitHub profiles
   - Google X-ray string for personal sites/portfolios

3. BUILD a talent map:
   - Tier 1: Direct competitors (same product/market)
   - Tier 2: Companies with similar tech stacks
   - Tier 3: Non-obvious companies where this talent hides

4. CREATE an ideal candidate persona:
   - What their LinkedIn headline probably says
   - What repos they contribute to on GitHub
   - What conferences they speak at
   - What motivates them to move

Always output in a structured format I can act on immediately.
When I share candidate profiles with you, evaluate them against my criteria and score them 1-5.
When I ask for outreach, write personalized messages under 100 words.

I'm looking for ML engineers who work with PyTorch and are based in the Bay Area. Generate GitHub search queries I can run in my browser.

Browser search (paste into github.com/search):
  language:python "pytorch" location:"San Francisco" type:users

Google X-ray for GitHub:
  site:github.com "pytorch" "machine learning" "San Francisco" -site:github.com/topics

Direct profile search:
  https://github.com/search?q=location%3A%22San+Francisco%22+language%3APython+pytorch&type=users

site:linkedin.com/in/ "machine learning engineer" "PyTorch" "San Francisco"

site:linkedin.com/in/ ("senior data engineer" OR "staff data engineer") "Spark" "Snowflake" "remote"

site:github.com "contributions" "pytorch" "tensorflow" "machine learning"

YOU: Here's a JD for a Senior ML Engineer at a Series B fintech startup 
in NYC. [paste JD]

Parse this and give me:
1. Structured criteria
2. LinkedIn Boolean (broad and targeted)
3. Google X-ray for LinkedIn
4. Google X-ray for GitHub
5. Top 15 companies where this person probably works today

CLAUDE: [gives you all of the above]

YOU: I found these 5 profiles. [paste LinkedIn URLs or summaries]
Score each one against the JD and rank them.

CLAUDE: [scores and ranks them with reasoning]

YOU: Write a personalized outreach message for the top 2 candidates.
Lead with what's impressive about THEM, not the job.

CLAUDE: [writes two personalized messages]

Create a Google Sheets template for a recruiting pipeline tracker. 
I need tabs for: Candidates, Requisitions, Submissions, and Activity Log.
Include columns, data validation rules, and conditional formatting 
for status tracking. Make it color-coded so I can see my pipeline at a glance.

URL: https://api.anthropic.com/v1/messages
Method: POST
Headers:
  x-api-key: [YOUR_API_KEY]
  anthropic-version: 2023-06-01
  content-type: application/json

Body:
{
  "model": "claude-sonnet-4-20250514",
  "max_tokens": 1024,
  "messages": [
    {
      "role": "user", 
      "content": "Write a personalized recruiting outreach message for [CANDIDATE NAME] who is a [TITLE] at [COMPANY]. I'm pitching a [ROLE] at [CLIENT]. Keep it under 100 words. Lead with something specific about their background."
    }
  ]
}

Parse this job description and extract:

1. MUST-HAVE skills (true dealbreakers, not wish list)
2. NICE-TO-HAVE skills (won't reject without)
3. Hidden requirements (implied but not stated)
4. Title variations (what this person calls themselves on LinkedIn)
5. Seniority signals (IC vs lead vs manager)
6. Comp estimate based on requirements + market
7. Top 10 companies where this exact person works today
8. Red flags (unrealistic combos, title/comp mismatch)

JD:
[PASTE JD HERE]

Generate search strings for this role:

Title: [ROLE]
Skills: [SKILLS]
Location: [LOCATION]
Experience: [YEARS]

I need:
1. LinkedIn Boolean (broad version, 200+ results)
2. LinkedIn Boolean (targeted version, 30-50 results)
3. Google X-ray for LinkedIn: site:linkedin.com/in/
4. Google X-ray for GitHub: site:github.com
5. Google X-ray for personal sites (exclude LinkedIn and GitHub)
6. Title variations I should also search

For each string, explain the logic so I can modify it.

Evaluate this candidate against the role we discussed.

Candidate info:
[PASTE LINKEDIN SUMMARY, RESUME, OR GITHUB PROFILE]

Score on these dimensions (1-5):
1. Technical skill match
2. Seniority alignment
3. Industry relevance
4. Career trajectory (trending up or down?)
5. Risk factors (gaps, hopping, title inflation)

Then give me:
- Overall verdict: SUBMIT / MAYBE / PASS
- Top 3 strengths to highlight to the client
- Top 3 risks to probe in a screen
- Suggested talking points for the candidate call

Write a cold outreach message for this candidate:

Name: [NAME]
Current role: [TITLE at COMPANY]
Something notable: [SPECIFIC DETAIL - project, patent, talk, open source work]
Role I'm pitching: [TITLE at CLIENT COMPANY]
Why this role is compelling: [2-3 selling points]

Rules:
- Under 100 words
- Lead with what impressed me about THEM
- No "exciting opportunity" or "I came across your profile"
- No corporate buzzwords
- End with a low-friction ask (15-min call, not a full interview)
- Sound like a human texting a colleague, not a recruiter template

I'm sourcing for [ROLE] with [KEY SKILLS]. Build me a talent map.

List 20 companies organized by:

TIER 1 - Direct competitors (same product/market):
[list companies + which team/org to target]

TIER 2 - Adjacent companies (similar tech stack):
[list companies + why their engineers would have these skills]

TIER 3 - Non-obvious sources:
[list companies nobody else is looking at + why this talent exists there]

For each company, tell me:
- Estimated team size for this function
- Recent news (layoffs, funding, product launches) that affects hiring
- Are they currently hiring for similar roles? (competition signal)

URL: https://api.openai.com/v1/chat/completions
Method: POST
Headers:
  Authorization: Bearer [YOUR_OPENAI_API_KEY]
  Content-Type: application/json

Body:
{
  "model": "gpt-4o",
  "messages": [
    {
      "role": "system",
      "content": "You are a recruiting sourcing assistant. When given a candidate and role, write personalized outreach under 100 words."
    },
    {
      "role": "user",
      "content": "Write outreach for [CANDIDATE] at [COMPANY] for my [ROLE] opening at [CLIENT]."
    }
  ]
}

# Install OpenClaw
npx openclaw@latest onboard

# It will walk you through:
# 1. Setting up a messaging channel (Telegram is easiest)
# 2. Connecting an LLM (Claude API or OpenAI API key)
# 3. Enabling web search (Exa API - free credits on signup)

# Resume parsing + candidate scoring + outreach drafting
npx playbooks add skill openclaw/skills --skill recruitment-automation

# Browser automation for LinkedIn/GitHub research
npx playbooks add skill TheSethRose/agent-browser

# Gmail + Google Calendar integration
openclaw skill install gog

# Web search for candidate research
openclaw skill install exa

# Recruiting Agent

You are a senior technical recruiter specializing in Data, ML, and AI roles.

## Your workflow
1. When given a job description, parse it into structured criteria
2. Search GitHub, LinkedIn (via browser), and the web for matching candidates
3. Score each candidate 1-5 against the requirements
4. Populate a Google Sheet with the shortlist
5. Draft personalized outreach emails for the top candidates

## Rules
- Never submit candidates below a 4.0 score
- Always verify current employment before outreach
- Keep outreach under 100 words
- Lead with what's impressive about the candidate, not the job
- Log every action for audit trail

## Daily routine
- 7:00 AM: Check for new job postings at target companies
- 9:00 AM: Brief me on yesterday's outreach responses
- 5:00 PM: Summary of candidates sourced today

# Monitor GitHub trending for ML engineers every morning
openclaw cron add --name "github-ml-monitor" \
  --schedule "0 8 * * *" \
  --prompt "Search GitHub trending Python repos tagged machine-learning. Find contributors in San Francisco with 50+ commits. Add promising profiles to my recruiting sheet."

# Check target company career pages daily
openclaw cron add --name "career-page-scan" \
  --schedule "0 9 * * *" \
  --prompt "Visit the careers pages for Anthropic, Scale AI, Databricks, and Snowflake. Report any new ML or Data Engineering roles posted in the last 24 hours."

# Weekly layoff alert
openclaw cron add --name "layoff-monitor" \
  --schedule "0 10 * * 1" \
  --prompt "Search for tech layoffs announced in the past week. Focus on companies in AI, data, and fintech. For each, estimate how many ML/Data engineers were affected."