Job Search Automation

Job Search Automation with Kimi WebBridge

This skill helps you search for any job role across multiple platforms using browser automation with Kimi WebBridge.

When to Use

• User asks "find jobs for [ROLE]"
• User wants to search multiple job boards in parallel
• User needs market intelligence on hiring for [SKILL/DOMAIN]
• User wants to track specific companies hiring for [ROLE]

Key Principle: User-Defined Search

Always ask the user what role/keywords they want to search for. The skill provides the methodology and platform patterns; the user provides the topic.

---

Platform Compatibility Matrix

| Platform | Works Well | Blockers | Best For | |----------|------------|----------|----------| | LinkedIn | ✅ Excellent | Login wall for details | Salary-filtered searches (f_SB2=200000), remote filter (f_WT=2), structured extraction | | Indeed (in.indeed.com) | ✅ Excellent | Pagination handling | Best for India-market remote roles. High-yield for specific tech stacks | | Glassdoor (.co.in) | ✅ Good | Must use direct job-listing URLs | Direct search result pages, "Remote only" filter, "Easy Apply" filter | | Naukri | ✅ Good | Pagination | India-focused, explicit experience/location filters (experience=0-3&location=remote) | | Dice | ✅ Good | Cookie consent dialog | Tech-focused, shows salary ranges, contract rates | | Wellfound | ❌ Blocked | DataDome CAPTCHA | Manual only | | Lever | ❌ Broken | Search URL 404s | — | | Greenhouse | ❌ Broken | Search URL 404s | — |

---

Effective Search Patterns (Template-Based)

LinkedIn (Best for High-Salary Remote)

https://www.linkedin.com/jobs/search/?keywords=[KEYWORDS]&location=[LOCATION]&f_WT=2&f_SB2=[SALARY]

• f_WT=2 = Remote
• f_SB2=200000 = $200K+ base salary filter
• Keywords: User-provided role terms

Indeed (Best for India-Market Remote)

https://in.indeed.com/jobs?q=[KEYWORDS]&location=Remote

• Use in.indeed.com for India market
• fromage=7 = last 7 days filter
• Keywords: User-provided role + tech stack terms

Dice

https://www.dice.com/jobs?q=[KEYWORDS]&location=Remote

• Shows contract rates and full-time salaries

Glassdoor (Use Direct Job URLs)

https://www.glassdoor.co.in/Job/[ROLE]-jobs-SRCH_KO0,17.htm?clickSource=searchBox

• Filter: "Remote only" button (@e24 in snapshot)

Naukri (India-Focused)

https://www.naukri.com/[ROLE]-jobs-in-india?k=[KEYWORDS]&experience=[0-3]&location=remote

• experience=0-3 = 0-3 years filter
• location=remote = remote filter

Wellfound (Manual Only - CAPTCHA Blocked)

https://wellfound.com/jobs?q=[KEYWORDS]&remote=true

• User must search manually

---

Tab Group Workflow (Kimi WebBridge)

macOS / Linux

# All tabs use same session name for grouping
session="[USER_DEFINED_SESSION]"
group_title="Job Search - [USER_DEFINED_ROLE]"

# Open N tabs with newTab:true (single group)
SEARCH_URLS=(
  "https://in.indeed.com/jobs?q=[KEYWORDS1]&location=Remote"
  "https://in.indeed.com/jobs?q=[KEYWORDS2]&location=Remote"
  "https://www.linkedin.com/jobs/search/?keywords=[KEYWORDS3]&location=India&f_WT=2"
  "https://www.naukri.com/[ROLE]-jobs-in-india?k=[KEYWORDS4]&experience=0-3&location=remote"
  # Add more as needed
)

for url in "${SEARCH_URLS[@]}"; do
  curl -s -X POST http://127.0.0.1:10086/command \
    -H 'Content-Type: application/json' \
    -d "{\"action\":\"navigate\",\"args\":{\"url\":\"$url\",\"newTab\":true,\"group_title\":\"$group_title\"},\"session\":\"$session\"}"
done

Windows (PowerShell)

$urls = @(
  "https://in.indeed.com/jobs?q=[KEYWORDS1]&location=Remote"
  "https://in.indeed.com/jobs?q=[KEYWORDS2]&location=Remote"
  "https://www.linkedin.com/jobs/search/?keywords=[KEYWORDS3]&location=India&f_WT=2"
  "https://www.naukri.com/[ROLE]-jobs-in-india?k=[KEYWORDS4]&experience=0-3&location=remote"
)

$session = "[USER_DEFINED_SESSION]"
$group_title = "Job Search - [USER_DEFINED_ROLE]"

foreach ($url in $urls) {
  $body = "{\"action\":\"navigate\",\"args\":{\"url\":\"$url\",\"newTab\":true,\"group_title\":\"$group_title\"},\"session\":\"$session\"}"
  $body | Set-Content -Path "$env:TEMP\kimi-nav.json" -Encoding UTF8
  curl.exe -s -X POST http://127.0.0.1:10086/command -H "Content-Type: application/json" --data-binary @"$env:TEMP\kimi-nav.json"
}

Extract from Each Tab

for url in "${SEARCH_URLS[@]}"; do
  curl -s -X POST http://127.0.0.1:10086/command \
    -H 'Content-Type: application/json' \
    -d "{\"action\":\"find_tab\",\"args\":{\"url\":\"$url\"},\"session\":\"$session\"}"
  
  curl -s -X POST http://127.0.0.1:10086/command \
    -H 'Content-Type: application/json' \
    -d "{\"action\":\"snapshot\",\"args\":{\"full\":true},\"session\":\"$session\"}"
  
  # Run platform-specific extraction via evaluate
done

# Cleanup
curl -s -X POST http://127.0.0.1:10086/command \
  -H 'Content-Type: application/json' \
  -d "{\"action\":\"close_session\",\"args\":{},\"session\":\"$session\"}"

---

Platform-Specific Extraction (via evaluate)

LinkedIn Job Cards

var jobs = document.querySelectorAll("[data-job-id]");
Array.from(jobs).slice(0, 20).map(el => ({
  title: el.querySelector("[data-job-title]")?.textContent?.trim() || "",
  company: el.querySelector("[data-company-name]")?.textContent?.trim() || "",
  location: el.querySelector("[data-job-location]")?.textContent?.trim() || "",
  link: el.querySelector("a[href*='/jobs/view/']")?.href || "",
  salary: el.querySelector("[data-salary]")?.textContent?.trim() || ""
}))

Indeed Job Cards (in.indeed.com)

// Use snapshot @e refs for job cards
// Extract: heading (title), StaticText (company, location), link (apply URL)

Glassdoor Job Cards

// Elements: heading (title), company logo/name, location, salary range, "Easy Apply" badge

Naukri Job Cards

// Snapshot shows: job title link, company, location, salary, experience, skills tags
// Filter URL params: experience=0-3&location=remote works well

---

Keyword Strategy

User Provides Role → Agent Generates Keywords

| User Input | Agent Should Search | |------------|---------------------| | "Python developer" | python developer remote, python engineer remote, django developer remote, fastapi developer remote | | "ML engineer" | machine learning engineer remote, ml engineer python remote, AI engineer remote, deep learning engineer remote | | "DevOps" | devops engineer remote, site reliability engineer remote, platform engineer remote, kubernetes engineer remote | | "Data scientist" | data scientist remote, data science python remote, analytics engineer remote, MLOps engineer remote |

Best Practice: Combine role + tech stack + remote

• [ROLE] remote [TECH_STACK]
• [ROLE] [TECH_STACK] remote python
• remote [ROLE] [KEY_SKILL]

---

Data Extraction Output Format

{
  "query": "[search query used]",
  "platform": "[platform name]",
  "date_searched": "2026-07-25",
  "results": [
    {
      "title": "Job Title",
      "company": "Company Name",
      "location": "City, Country / Remote",
      "experience": "X+ years",
      "salary": "$X - $Y (if mentioned)",
      "skills": ["skill1", "skill2"],
      "apply_link": "URL or 'DM' or 'See comments'",
      "post_date": "Jul 2 / 19 hours ago",
      "job_id": "[platform-specific ID]"
    }
  ]
}

---

Pitfalls & Gotchas

1. LinkedIn blocks automation — event.isTrusted errors. Manual interaction needed for CAPTCHAs/login walls. Add delays between requests.
2. Indeed pagination — Handle start=10, start=20 etc.
3. Glassdoor login wall — Deep scroll triggers login. Capture first page only.
4. Naukri rate limits — Add delays, respect robots.txt
5. Wellfound CAPTCHA — Cannot automate. Manual only.
6. Dice cookie consent — Handle dialog before extraction.

---

References

• job-search-automation/references/search-urls.md — Search URL templates
• job-search-automation/references/extraction-scripts.md — Platform-specific JS
• job-search-automation/references/platform-quirks.md — Platform-specific handling

---

Quick Start for Agent

When user says "Find jobs for [ROLE]":

1. Clarify (if needed):

• "What specific role/keywords should I search?"
• "Any specific tech stack? (e.g., Python, React, AWS)"
• "Location preference? (Remote, India, US, EU, etc.)"
• "Salary minimum?"
• "Experience level?"

1. Build search URLs using templates above with user's keywords

1. Open 8-10 tabs in single tab group via Kimi WebBridge

1. Extract & synthesize results across platforms

1. Output structured data + apply links + company names for follow-up research

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: sloemo01
• Source: sloemo01/hermes-skills-bundle
• License: MIT

Install and usage instructions live in the source repository linked above.