jobcli

README

🧑‍💻 JobCLI - Jobs in your terminal

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

Features

• Concurrent scraping across LinkedIn, Indeed, Glassdoor, ZipRecruiter, Google Jobs, 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

Requirements

• Go 1.25

Installation

Homebrew (macOS/Linux)

brew install jimezsa/tap/jobcli

Windows

1. Download the latest .zip for your architecture from the Releases page:
   • jobcli_<version>_windows_amd64.zip for 64-bit Intel/AMD
   • jobcli_<version>_windows_arm64.zip for ARM64
2. Extract the .zip file
3. Move jobcli.exe to a directory in your PATH, or run it directly:

.\jobcli.exe --help

Build from source

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

Run:

jobcli --help

Quick Start

# show overview and command list
jobcli

# search software engineer roles in Munich, Germany
jobcli search "software engineer" --location "Munich, Germany"  --limit 100

# search a single site last 48 hours
jobcli linkedin "chemical engineer" --location "Munich, Germany"  --limit 10 --hours 48

# search a single site
jobcli stepstone "hardware engineer" --location "Munich, Germany"  --limit 100

# output only unseen jobs using a seen-history JSON
jobcli search "software engineer" --location "Munich, Germany" --limit 30 \
  --seen jobs_seen.json --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"

Commands

• jobcli version
• jobcli config init
• jobcli config path
• jobcli search <query> [--location L] [--sites S] [--limit N] [--offset N]
• jobcli linkedin <query> ...
• jobcli indeed <query> ...
• jobcli glassdoor <query> ...
• jobcli ziprecruiter <query> ...
• jobcli google <query> ...
• jobcli stepstone <query> ...
• 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

Output Formats

• 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

Flags

Global flags:

• --color=auto|always|never
• --json
• --plain
• --verbose
• --version

Search flags:

• --location
• --sites (comma-separated list; default all)
• --limit
• --offset
• --job-type=fulltime|parttime|contract|internship
• --hours
• --country
• --format=csv|json|md
• --links=short|full
• --output (aliases: --out, --file) (write the primary output to a file)
• --proxies (comma-separated URLs)
• --seen (path to seen jobs JSON history)
• --new-only (output only unseen jobs; requires --seen)
• --new-out (also write unseen jobs (A - B) to a JSON file; requires --seen)
• --seen-update (update --seen by merging in newly discovered unseen jobs after the search completes; requires --seen)

Seen flags:

• --stats (print diff/merge stats to stdout)

Seen Jobs Workflow

Use this when you want only fresh jobs in recurring runs.

# 1) scrape and keep only unseen jobs (C = A - B)
jobcli search "hardware engineer" --location "Munich, Germany" --limit 30 \
  --seen jobs_seen.json --new-only --json --output jobs_new.json

# If you want to auto-mark new jobs as "seen" in the same run (no separate
# `jobcli seen update` step), add --seen-update:
# jobcli search "hardware engineer" --location "Munich, Germany" --limit 30 \
#   --seen jobs_seen.json --new-only --seen-update --json --output jobs_new.json

# If you want to keep the primary output as "all jobs" (table/CSV/etc) but still
# write unseen jobs to JSON, use --new-out (no --new-only needed):
# jobcli search "hardware engineer" --location "Munich, Germany" --limit 30 \
#   --seen jobs_seen.json --format csv --output jobs_all.csv --new-out jobs_new.json

# 2) (optional) rank/review jobs_new.json with your own tooling

# 3) persist accepted/new jobs back into seen history
jobcli seen update --seen jobs_seen.json --input jobs_new.json --out jobs_seen.json --stats

Config

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

Proxy Checking

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

AI Agent Skills

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

• SKILL-cv-summary.md — reads a PDF CV, extracts an anonymous persona summary and 10 search keywords, and saves them to CVSUMMARY.md. Run once or whenever you update your CV.
• SKILL-jobcli-ranking.md — reads CVSUMMARY.md, runs jobcli searches for each keyword, deduplicates results, ranks every listing 0–1 against your persona, and presents a scored table. Designed for daily use.

Notes

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

Inspiration

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

License

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