job-market-intel

Job Market Intel — Live Salary & Demand Data Before You Negotiate

You have access to the TinyFish CLI (tinyfish), a tool that runs browser automations from the terminal using natural language goals. This skill deploys agents across LinkedIn, Indeed, and Glassdoor in parallel, navigates login walls and search forms, and synthesizes live job listings into a single market snapshot.

---

Pre-flight Check (REQUIRED)

Before making any TinyFish call, always run BOTH checks:

1. CLI installed?

PowerShell:

Get-Command tinyfish; tinyfish --version

bash/zsh:

which tinyfish && tinyfish --version || echo "TINYFISH_CLI_NOT_INSTALLED"

If not installed, stop and tell the user:

Install the TinyFish CLI: npm install -g @tiny-fish/cli

2. Authenticated?

tinyfish auth status

If not authenticated, stop and tell the user:

You need a TinyFish API key. Get one at: https://agent.tinyfish.ai/api-keys

Then authenticate:

Option 1 — CLI login (interactive):

tinyfish auth login

Option 2 — PowerShell (current session only):

$env:TINYFISH_API_KEY="your-api-key-here"

Option 3 — PowerShell (persist across sessions):

[System.Environment]::SetEnvironmentVariable("TINYFISH_API_KEY", "your-api-key-here", "User")

Then close and reopen PowerShell for it to take effect.

Option 4 — bash/zsh (Mac/Linux):

export TINYFISH_API_KEY="your-api-key-here"

Option 5 — Claude Code settings: Add to ~/.claude/settings.local.json:

{
  "env": {
    "TINYFISH_API_KEY": "your-api-key-here"
  }
}

Do NOT proceed until both checks pass.

---

What This Skill Does

Given a job title and tech stack (e.g. "Senior Backend Engineer, Go + Kubernetes"), this skill:

1. Searches LinkedIn Jobs for matching postings, extracting salary ranges, seniority levels, and company names
2. Searches Indeed for the same role, pulling compensation data and required skills from job descriptions
3. Searches Glassdoor for salary data, company ratings, and remote/hybrid/onsite ratios

It then synthesizes all three sources into a structured market snapshot: what the role actually pays, which skills show up most, who is hiring, and whether remote is realistic.

---

Core Command

tinyfish agent run --url <url> "<goal>"

Flags

---

Step-by-Step Workflow

Inputs Required

Before running, collect from the user:

• Job title — e.g. Senior Backend Engineer, ML Engineer, Staff iOS Developer
• Tech stack — e.g. Go Kubernetes, Python PyTorch, Swift SwiftUI
• Location (optional) — e.g. San Francisco, Remote, London

Derive URL-encoded search strings from these inputs for each platform.

---

Step 1 — Search LinkedIn Jobs

tinyfish agent run --sync \
  --url "https://www.linkedin.com/jobs/search/?keywords=<title+stack>&location=<location>&f_TPR=r604800" \
  "Extract the top 10 job listings as JSON: [{\"title\": str, \"company\": str, \"location\": str, \"salary_range\": str, \"posted_date\": str, \"remote_type\": str, \"url\": str}]. If salary is not shown, set salary_range to null. The f_TPR=r604800 filter means postings from the last 7 days — note the recency in each result."

Example for Senior Backend Engineer, Go + Kubernetes:

tinyfish agent run --sync \
  --url "https://www.linkedin.com/jobs/search/?keywords=Senior+Backend+Engineer+Go+Kubernetes&location=Remote&f_TPR=r604800" \
  "Extract the top 10 job listings as JSON: [{\"title\": str, \"company\": str, \"location\": str, \"salary_range\": str, \"posted_date\": str, \"remote_type\": str, \"url\": str}]. Set salary_range to null if not listed."

---

Step 2 — Search Indeed

tinyfish agent run --sync \
  --url "https://www.indeed.com/jobs?q=<title+stack>&l=<location>&fromage=7&sort=date" \
  "Extract the top 10 job listings as JSON: [{\"title\": str, \"company\": str, \"location\": str, \"salary_range\": str, \"required_skills\": [str], \"posted_date\": str, \"remote_type\": str, \"url\": str}]. Pull required_skills from the snippet or description preview shown in results. Set salary_range to null if not listed."

Example:

tinyfish agent run --sync \
  --url "https://www.indeed.com/jobs?q=Senior+Backend+Engineer+Go+Kubernetes&l=Remote&fromage=7&sort=date" \
  "Extract the top 10 job listings as JSON: [{\"title\": str, \"company\": str, \"location\": str, \"salary_range\": str, \"required_skills\": [str], \"posted_date\": str, \"remote_type\": str, \"url\": str}]. Pull required_skills from visible description snippets."

---

Step 3 — Search Glassdoor

tinyfish agent run --sync \
  --url "https://www.glassdoor.com/Job/jobs.htm?sc.keyword=<title+stack>&locT=N&locId=1&jobType=all&fromAge=7" \
  "Extract the top 10 job listings as JSON: [{\"title\": str, \"company\": str, \"company_rating\": str, \"salary_range\": str, \"location\": str, \"remote_type\": str, \"posted_date\": str, \"url\": str}]. Include company_rating (out of 5) if shown. Set salary_range to null if not listed."

Example:

tinyfish agent run --sync \
  --url "https://www.glassdoor.com/Job/jobs.htm?sc.keyword=Senior+Backend+Engineer+Go+Kubernetes&locT=N&locId=1&jobType=all&fromAge=7" \
  "Extract the top 10 job listings as JSON: [{\"title\": str, \"company\": str, \"company_rating\": str, \"salary_range\": str, \"location\": str, \"remote_type\": str, \"posted_date\": str, \"url\": str}]."

---

Parallel Execution

All three steps are independent — run them simultaneously. Do NOT wait for LinkedIn before starting Indeed or Glassdoor.

bash/zsh — Parallel calls (Mac/Linux):

tinyfish agent run --sync \
  --url "https://www.linkedin.com/jobs/search/?keywords=<title+stack>&location=<location>&f_TPR=r604800" \
  "Extract top 10 listings as JSON: [{\"title\": str, \"company\": str, \"location\": str, \"salary_range\": str, \"posted_date\": str, \"remote_type\": str, \"url\": str}]" \
  > /tmp/linkedin_results.json &

tinyfish agent run --sync \
  --url "https://www.indeed.com/jobs?q=<title+stack>&l=<location>&fromage=7&sort=date" \
  "Extract top 10 listings as JSON: [{\"title\": str, \"company\": str, \"location\": str, \"salary_range\": str, \"required_skills\": [str], \"posted_date\": str, \"remote_type\": str, \"url\": str}]" \
  > /tmp/indeed_results.json &

tinyfish agent run --sync \
  --url "https://www.glassdoor.com/Job/jobs.htm?sc.keyword=<title+stack>&locT=N&locId=1&jobType=all&fromAge=7" \
  "Extract top 10 listings as JSON: [{\"title\": str, \"company\": str, \"company_rating\": str, \"salary_range\": str, \"location\": str, \"remote_type\": str, \"posted_date\": str, \"url\": str}]" \
  > /tmp/glassdoor_results.json &

wait

PowerShell — Parallel calls using Start-Job / Wait-Job:

$linkedinJob = Start-Job {
    tinyfish agent run --sync `
      --url "https://www.linkedin.com/jobs/search/?keywords=<title+stack>&location=<location>&f_TPR=r604800" `
      "Extract top 10 listings as JSON: [{`"title`": str, `"company`": str, `"location`": str, `"salary_range`": str, `"posted_date`": str, `"remote_type`": str, `"url`": str}]"
}

$indeedJob = Start-Job {
    tinyfish agent run --sync `
      --url "https://www.indeed.com/jobs?q=<title+stack>&l=<location>&fromage=7&sort=date" `
      "Extract top 10 listings as JSON: [{`"title`": str, `"company`": str, `"location`": str, `"salary_range`": str, `"required_skills`": [str], `"posted_date`": str, `"remote_type`": str, `"url`": str}]"
}

$glassdoorJob = Start-Job {
    tinyfish agent run --sync `
      --url "https://www.glassdoor.com/Job/jobs.htm?sc.keyword=<title+stack>&locT=N&locId=1&jobType=all&fromAge=7" `
      "Extract top 10 listings as JSON: [{`"title`": str, `"company`": str, `"company_rating`": str, `"salary_range`": str, `"location`": str, `"remote_type`": str, `"posted_date`": str, `"url`": str}]"
}

Wait-Job $linkedinJob, $indeedJob, $glassdoorJob | Out-Null

$linkedinResults  = Receive-Job $linkedinJob
$indeedResults    = Receive-Job $indeedJob
$glassdoorResults = Receive-Job $glassdoorJob

Remove-Job $linkedinJob, $indeedJob, $glassdoorJob

Bad — Sequential calls (avoid):

# Don't do this — three times slower, same results
tinyfish agent run --url "https://linkedin.com/..." "...also check Indeed and Glassdoor..."

Each platform is its own call. Always run in parallel.

---

Step 4 — Synthesize Into a Market Snapshot

Once all three sources return results, synthesize findings into this structure:

## Job Market Snapshot: <Role> — <Stack>
As of: <current date> | Sources: LinkedIn · Indeed · Glassdoor

### Salary Range
- Low end:   
lt;X>k   (entry/mid, from listings with explicit salary data)
- Midpoint:  
lt;X>k   (most common range across listings)
- High end:  
lt;X>k   (senior/staff/principal listings)
- Sample size: <N> listings with salary data out of <total> scraped

### Top Hiring Companies
| Company | Platform | Rating | Remote? |
|---------|----------|--------|---------|
| ...     | ...      | ...    | ...     |

### Most Required Skills
(ranked by frequency across all listings)
1. <skill> — seen in X/30 listings
2. <skill> — seen in X/30 listings
...

### Remote Ratio
- Remote:        XX%
- Hybrid:        XX%
- Onsite only:   XX%

### Demand Signal
- Total listings found this week: ~<N>
- Platform with most postings: <LinkedIn / Indeed / Glassdoor>
- Trend note: <any notable pattern — e.g. most postings are contract, or salary listed in <20% of roles>

### Raw Listings Sample
<3–5 representative listings with title, company, salary, and URL>

### Negotiation Takeaway
<2–3 sentences: what this data means in a negotiation context — floor, target, and stretch numbers based on the listings>

Only use data returned by TinyFish. Do not estimate or fill in salary figures not present in the results. If salary data is sparse (fewer than 5 listings show it), say so explicitly and note which platform surfaces it most.

---

Keyword Strategy

Search quality depends on how you phrase the query. Derive 2–3 variants from the user's input:

If the first pass returns fewer than 10 total listings across all platforms, run a second parallel pass with a variant keyword set.

---

Handling Login Walls

LinkedIn and Glassdoor may prompt for login. TinyFish handles this automatically when authenticated sessions are stored. If an agent returns a login page instead of results:

1. Run tinyfish auth status to verify your session
2. Re-authenticate with tinyfish auth login and select the affected platform
3. Re-run the failed step individually before re-joining the parallel flow

Indeed typically does not require login for search result extraction.

---

Managing Runs

# List recent runs
tinyfish agent run list

# Get a specific run by ID
tinyfish agent run get <run_id>

# Cancel a running automation
tinyfish agent run cancel <run_id>

---

Output Format

The CLI streams data: {...} SSE lines by default. The final result is the event where type == "COMPLETE" and status == "COMPLETED" — extracted data is in the resultJson field. Read output directly; no custom parsing script needed.