Home

AutoApply Wiki

AutoApply is an AI-powered job application automation tool that searches for jobs, generates tailored resumes and cover letters, and applies automatically across multiple ATS platforms.

Current Version: v1.9.0 | Production Readiness: 10.0/10 | Test Coverage: 97% (bot/core/config/db)

---

Quick Links

Section	Description
Getting Started	Installation, setup wizard, first run
Architecture Overview	System design, components, data flow
API Reference	All REST endpoints with request/response examples
Configuration Guide	Settings, LLM providers, search criteria, scheduling
Bot Operations	How the bot works, apply modes, ATS support
Development Guide	Local setup, testing, CI/CD, contributing
Internationalization	i18n system, adding new languages
Distribution & Packaging	Building installers, release workflow
Security	Auth, headers, rate limiting, threat model
Changelog	Version history from v1.0.0 to v1.9.0
Roadmap	Upcoming features and enhancements

---

Project Structure

AutoApply/
├── app.py                  # Flask app factory + middleware
├── app_state.py            # Shared mutable state (thread-safe)
├── run.py                  # Entry point (logging, port detection, gevent)
├── config/settings.py      # Pydantic config models
├── core/
│   ├── ai_engine.py        # Multi-provider LLM API
│   ├── filter.py           # Job scoring + ATS detection
│   ├── resume_renderer.py  # PDF generation (ReportLab)
│   ├── scheduler.py        # Time-based bot scheduling
│   └── i18n.py             # Backend translation system
├── db/
│   ├── database.py         # SQLite operations (WAL mode)
│   └── models.py           # Pydantic data models
├── bot/
│   ├── bot.py              # Main loop: search → filter → generate → apply
│   ├── browser.py          # BrowserManager (Playwright persistent context)
│   ├── state.py            # Bot state machine
│   ├── search/             # LinkedIn, Indeed searchers
│   └── apply/              # 6 appliers: LinkedIn, Indeed, Greenhouse, Lever, Workday, Ashby
├── routes/                 # 7 Flask Blueprints (bot, applications, config, profile, analytics, login, lifecycle)
├── static/
│   ├── css/main.css        # All styles
│   ├── js/                 # 17 ES modules (no build step)
│   └── locales/en.json     # 383 translation keys
├── templates/index.html    # SPA shell with data-i18n attributes
├── electron/               # Desktop shell (main.js, python-backend.js, tray, build scripts)
├── tests/                  # 738 tests across 27 files
└── .github/workflows/      # CI (lint+test+security) + Release (3-platform builds)

---

Architecture Decisions

ADR	Decision	Rationale
ADR-005	Electron wrapping Flask	Desktop app with Python backend
ADR-006	Separate Chromium for Playwright	Persistent browser contexts incompatible with Electron's Chromium
ADR-008	Port auto-detection (5000-5010)	Avoid conflicts on common ports
ADR-009	Multi-provider LLM via direct HTTP	No SDK dependency, supports 4 providers
ADR-010	ReportLab for PDF	ATS-safe resume rendering
ADR-011	Fallback templates	Works without AI configured
ADR-014	Flask Blueprint architecture	7 blueprints, shared state module
ADR-017	Vanilla ES modules	No bundler, native browser support
ADR-018	Python bundling strategy	Windows embeddable + python-build-standalone
ADR-019	Programmatic icon generation	canvas + png2icons, no manual assets
ADR-020	CI release on v* tags	GitHub Actions → GitHub Releases

---

Requirements Traceability

Category	Total	Covered	Partial	Missing
Functional (FR-001–082)	79	66	13	0
Quick Wins (QW-1–5)	5	5	0	0
Medium Effort (ME-1–9)	9	9	0	0
Deferred (D-5–7)	3	3	0	0
Large Effort (LE-1–3)	3	3	0	0
Distribution (DIST-01–09)	9	9	0	0
Total	108	95	13	0

The 13 partial items are Electron E2E and frontend DOM rendering — cannot be tested with pytest.