What It Is

This system is a production Flask API for Google account login automation at scale. It orchestrates AdsPower anti-detect browser profiles, location-aligned proxies, and Xano as the system of record running single and batch logins, tracking step-level attempt logs, validating signed-out sessions, and collecting browser cookies through configured post-login browsing flows.

The architecture treats browser farms as an operations platform: parallel workers, profile reuse, evidence capture, and Grafana-backed visibility not one-off scripts.

The Problem We Solved

Managing thousands of browser-backed accounts manually does not scale:

Each account needs an isolated fingerprint (AdsPower profile) and a consistent proxy strategy
Login flows fail in many ways (CAPTCHA, challenges, terms, recovery) and need structured retry metadata
Operators lack a single API to run batches, inspect runs, and see which step failed
Session drift (signed-out Gmail) wastes profiles until detected and reclaimed
Cookie state after login must be harvested and stored for downstream tools

The service centralizes login orchestration, persistence, and maintenance so engineering and operations share one control plane.

What We Work On

Single and batch login endpoints, parallel workers (MAX_PARALLEL_ACCOUNTS), and step-level account_login_attempts with run-level aggregates.

Profile & proxy management

AdsPower profile lifecycle, proxy assignment with geo alignment, fallback modes, and profile cleanup APIs.

Session maintenance

Signed-out detection, scheduled validity checks, session reopen with backoff, and resource reclamation.

Post-login top-sites browsing and CDP cookie capture persisted to Xano for accounts that completed successfully.

Resilience & safety

CAPTCHA integrations (2Captcha, Anti-Captcha, CapMonster), recovery and terms handlers, evidence screenshots, and account ban/unban controls.

Observability & ops

Flasgger API docs, Grafana dashboards (Xano/Infinity and optional MySQL sources), APScheduler crons, and Docker/PM2 deployment patterns.

How It Works (In Simple Terms)

Configure: Accounts, proxies, and AdsPower profiles are registered in Xano.
Assign: The service picks or creates a profile and aligns a proxy to the account’s location.
Automate: Selenium drives the browser inside AdsPower for Google sign-in and verification steps.
Record: Each step and run is logged; failures store challenge and cooldown metadata.
Maintain: Cron jobs check session/cookie validity, reopen sessions, and run cookie-farming flows.

The API is the orchestration layer; AdsPower owns browser isolation and Xano owns durable state.

Key Outcomes

Scalable throughput: Parallel login workers across a large account pool.
Fingerprint isolation: Per-account AdsPower profiles instead of shared browser state.
Operational visibility: Step logs, run metrics, and Grafana views for incidents.
Automated hygiene: Signed-out checks and cookie-farming pipelines reduce manual rework.
Production hardening: Evidence capture, ban controls, and documented runbooks for operators.

Technologies & Approaches We Used

Area	What we used	Why it matters
API	Flask, Gunicorn, Flasgger	REST control plane with Swagger for operators
Automation	Selenium 4, AdsPower local API	Real browser flows with anti-detect profiles
Data	Xano REST repositories	Central SoT for accounts, runs, proxies, cookies
Scheduling	APScheduler	Session validity, cookie checks, maintenance crons
Validation	Pydantic, python-dotenv	Typed config and safer deployments
CAPTCHA	2Captcha, Anti-Captcha, CapMonster	Automated recovery on challenge steps
Monitoring	Grafana, Docker Compose	Dashboards and local ops stack
Deploy	PM2, Make	Repeatable prod/dev process management

Approach in practice: We kept orchestration in Python, state in Xano, and browsers in AdsPower so each layer can scale independently. Parallelism, geo-matched proxies, and structured attempt logs make failures diagnosable without replaying full manual sessions.

Who It's For

Internal platform teams managing large browser-account fleets
Operators running batch logins, session checks, and cookie collection
Engineers integrating anti-detect automation with a centralized data backend