ApplySemPatria (jobsearch CLI)

Local-first job discovery and application tooling (boards, LinkedIn posts, email/DM/form apply channels, applications UI).

Install and run everything with make from a single clone. Your data (tracks/, registry/, state/) stays on your machine and is not uploaded.

Requirements

• Python 3.11+
• Make (macOS/Linux — make is preinstalled on most systems)
• Git (full clone — the CLI expects flows/, ui/, examples/ next to scripts/)
• macOS or Linux recommended (browser apply is tested on macOS)

Quick start

git clone https://github.com/ProgramadoresSemPatria/ApplySemPatria.git job-search
cd job-search

make quickstart
make onboarding
make ui

That’s it for setup — no pip, no source .venv/bin/activate. Run make help for everything else (discover, doctor, reset-data, Docker, tests).

When you need LinkedIn or form apply in the browser: make browser once.

Daily commands

Goal	Command
Discover jobs	make discover or make discover SINCE=14d TRACK=ai-engineer
Open dashboard	make ui
Health check	make doctor
List tracks	make tracks
Reinstall deps after git pull	make upgrade then make doctor
Install Chromium via CLI	make install-cli

Reset local data (bugs / first-run replay)

Backs up beside the repo by default, then wipes personal data:

make reset-data              # preview — no changes
make reset-data CONFIRM=1    # backup + wipe tracks/, registry/, state/, secrets/, runs/, logs/
make fresh-start CONFIRM=1   # reset + bootstrap
make onboarding

Options: BACKUP=0, RESET_LINKEDIN=0 (keep LinkedIn cookies), RESET_BROWSER=1 (remove patchright-profile/).

Keeping your install stable

Step	Why
Use one clone and run make … from that folder	Paths, flows/, and state/ are relative to the repo root
Stay on main or a release tag (git checkout v0.1.0)	Tags are supported snapshots
After git pull: make upgrade and make doctor	Aligns Python packages and surfaces missing setup
Never commit tracks/, secrets/, state/, registry/	Gitignored — data stays local

Pin a release when we tag them:

git fetch --tags
git checkout v0.1.0
make install
make doctor

What's in git vs local

In git	Local only (.gitignore)
scripts/, tests/, ui/, flows/, playbooks/	tracks/ — your profile & preferences
examples/tracks/ — sanitized templates	registry/ — discovered jobs
tracks.json — track manifest	runs/ — generated tables & audit output
domain-blacklist.json — shared scam domains	state/ — apply progress, Gmail token
Makefile, scripts/reset_local_data.sh	secrets/ — credentials

Cursor / agent-assisted apply (optional)

The CLI runs deterministic steps (discover, table, Playwright connect flows, email).
Form apply and fuzzy judgment are designed to work with an IDE agent (e.g. Cursor + skills) or optional OPENAI_API_KEY for LinkedIn post intent (llm_intent_classify_enabled in track config). No agent is bundled in the pip package.

See playbooks/ for channel-specific notes (Gmail OAuth, LinkedIn session, Himalayas, etc.).

Docker (isolated install + e2e)

Same project, no local tracks/ or secrets — data only inside the container:

make docker-build
make docker-e2e          # CI-style smoke
make docker-ui           # http://127.0.0.1:8765/
make docker-onboard      # interactive onboarding in container

GitHub Actions runs docker-e2e.yml on every push to main.

Testing (contributors)

make test              # unit + browser
make test-unit
make test-coverage
make test-stable

Scenario catalog: docs/testing/SCENARIOS.md
CI strategy: docs/testing/STRATEGY.md

Publishing releases (maintainers)

1. Bump version in pyproject.toml, commit, tag: git tag v0.1.0 && git push origin v0.1.0
2. CI on main should stay green (make test-unit)
3. Users: git clone at that tag, then make install

PyPI-only pip install jobsearch-cli without a clone is not supported yet (data files and tracks/ layout require the full repo).

Manual install (no Make)

python3 -m venv .venv && source .venv/bin/activate
pip install -e ".[browser,gmail]"
PLAYWRIGHT_BROWSERS_PATH="$HOME/.linkedin-mcp/patchright-browsers" patchright install chromium
bash scripts/bootstrap_local.sh ai-engineer
jobsearch onboarding --track ai-engineer

Legacy: pip install -r requirements.txt and python scripts/jobsearch.py doctor.