Fast, single-binary job aggregation CLI written in Go. Scrapes multiple sites in parallel and exports results to table, CSV, TSV, JSON, or Markdown.

• Concurrent scraping across LinkedIn, Indeed, Glassdoor, ZipRecruiter, and Stepstone
• TLS fingerprinting via tls-client to reduce blocking
• Proxy rotation with temporary bans on 403/429 responses
• Seen-jobs workflow with JSON diff/update commands to avoid reprocessing old listings
• Human-friendly tables or machine-friendly exports
• Config + proxies stored in the user config directory

• Go 1.25

Choose one method:

• macOS/Linux (Homebrew): brew install jimezsa/tap/jobcli
• Windows: download the latest .zip from Releases, extract, and run jobcli.exe (or add it to PATH)
• Build from source:

git clone https://github.com/jimezsa/jobcli
cd jobcli
make
./jobcli --help

# show overview and command list
jobcli

# search software engineer roles in Munich, Germany
jobcli search "software engineer" --location "Munich, Germany"  --limit 20 --hours 48

# search multiple queries in one run (comma-separated)
jobcli search "software engineer,hardware engineer,data scientist" --location "Munich, Germany" --limit 100 --hours 48

# load queries from JSON file
jobcli search --query-file queries.json --location "Munich, Germany" --limit 200 --hours 48

# search a single site last 48 hours
jobcli linkedin "chemical engineer" --location "Munich, Germany"  --limit 20 --hours 48

# output only unseen jobs using a seen-history JSON , update seen jobs (jobs_seen.json)
jobcli search "software engineer" --location "Munich, Germany" --limit 30 --hours 48 \
  --seen jobs_seen.json --seen-update --new-only --json --output jobs_new.json

# update seen-history after reviewing/ranking new jobs
jobcli seen update --seen jobs_seen.json --input jobs_new.json --out jobs_seen.json --stats

# avoid 403s by narrowing sites or providing proxies
jobcli search "software engineer" --sites linkedin --location "Munich, Germany" --country de --limit 10
jobcli search "software engineer" --location "Munich, Germany" --country de --proxies "http://user:pass@host:port,http://host2:port"

Recommendation: do not use LinkedIn scraping too intensively; prefer --limit 20 instead of very high values like --limit 200.

Example queries.json:

{
  "job_titles": ["software engineer", "hardware engineer", "data scientist"]
}

Also supported:

["software engineer", "hardware engineer", "data scientist"]

Full profile (queries + defaults):

{
  "job_titles": ["software engineer", "hardware engineer"],
  "search_options": {
    "location": "Munich, Germany",
    "limit": 5,
    "hours": 28,
    "seen": "jobs_seen.json",
    "seen_update": true,
    "new_only": true,
    "output": "jobs_new.json"
  },
  "global_options": {
    "json": true
  }
}

Then run only:

jobcli search --query-file queries.json

The skills/ directory contains ready-made skill files that let AI coding agents (OpenClaw, Cursor, Codex, etc.) use JobCLI on your behalf:

• skills/jobcli-cv-summary/SKILL.md (jobcli-cv-summary) — reads one or more CV PDFs and writes a privacy-safe persona/search profile to profiles/<user_id>/persona_querie.json (compatible with --query-file).
• skills/jobcli-job-search/SKILL.md (jobcli-job-search) — reads profiles/<user_id>/persona_querie.json, runs unseen-job searches via --query-file, and keeps only YES matches above the configured confidence threshold.
• skills/pirate-motivator/SKILL.md (pirate-motivator) — generates short pirate-style motivational audio for job-hunt encouragement.

• jobcli version
• jobcli config init
• jobcli config path
• jobcli search [<query>] [--query-file queries.json] [--location L] [--sites S] [--limit N] [--offset N]
• jobcli linkedin [<query>] [--query-file queries.json] ...
• jobcli indeed [<query>] [--query-file queries.json] ...
• jobcli glassdoor [<query>] [--query-file queries.json] ...
• jobcli ziprecruiter [<query>] [--query-file queries.json] ...
• jobcli stepstone [<query>] [--query-file queries.json] ...
• jobcli seen diff --new A.json --seen B.json --out C.json [--stats]
• jobcli seen update --seen B.json --input C.json --out B.json [--stats]
• jobcli proxies check

• Default: table when stdout is a TTY, CSV otherwise (columns: site/title/company/url; URL is blue)
• --json: JSON array
• --plain: TSV
• --format=csv|json|md: explicit format override

Global flags:

Search and site flags (search, linkedin, indeed, glassdoor, ziprecruiter, stepstone):

Seen command flags:

Proxy command flags:

Notes:

• search supports comma-separated positional query lists (max 10), e.g. "backend,platform,sre".
• --query-file accepts ["backend","platform"], {"job_titles":["backend","platform"]}, or a full profile object with optional search_options and global_options.
• Option precedence is: explicit CLI flags > query-file defaults > env/config defaults.
• Positional and file queries can be combined; positional entries are applied first, then deduped case-insensitively.

Config directory:

$(os.UserConfigDir())/jobcli/

Files:

• config.json
• proxies.txt
• cookies.json (optional)

Environment variables:

• JOBCLI_COLOR=auto|always|never
• JOBCLI_JSON=1
• JOBCLI_VERBOSE=1
• JOBCLI_PROXIES=...
• JOBCLI_DEFAULT_LOCATION="New York, NY"
• JOBCLI_DEFAULT_COUNTRY="usa"
• JOBCLI_DEFAULT_LIMIT=20

jobcli proxies check --target "https://www.google.com" --timeout 15

• Scrapers are best-effort and may require selector updates as sites change.
• Heavy usage may require rotating proxies.

• JobSpy: https://github.com/speedyapply/JobSpy
• gogcli: https://github.com/steipete/gogcli

This project is licensed under the MIT License - see the LICENSE file for details.