Workdir
`SCOUT_WORKDIR` is where Scout writes run artifacts: records, source pages, blocked pages, validation, reports, raw evidence, and screenshots.
Scout Quickstart
Scout is safest to test on your own machine first: your workdir, your artifacts, your optional keys, and your browser capture path.
Path 01
python3 -m venv .venv
source .venv/bin/activate
pip install "git+https://github.com/arijitchowdhury80/scout.git@codex/scout-platform-foundation"
playwright install chromium
export SCOUT_WORKDIR="$PWD/scout-runs"
scout serve
`SCOUT_WORKDIR` is where Scout writes run artifacts: records, source pages, blocked pages, validation, reports, raw evidence, and screenshots.
Open `http://127.0.0.1:8421/health` after `scout serve` starts. The launch website is served from `/`. Local Swagger API docs remain at `/docs`; hosted deployments keep Swagger private.
`pip install scout-web` comes after the registry publishing and license gates close. Until then, beta installs use the verified launch branch.
Path 02
docker compose -f docker/docker-compose.yml up -d
curl http://127.0.0.1:8421/health
Mount a host volume for Scout runs so artifacts survive container restarts.
Docker is best for service-style scraping. User-browser capture remains a local-machine workflow because it can touch your active session.
First useful run
Start with a public page you are allowed to access.
Use CLI or HTTP API. Keep live hard-site testing small and intentional.
Open `records.json`, `source_pages.json`, `blocked_pages.json`, and `extraction_report.md`.
Readiness check
The readiness checker summarizes the current truth: private beta is ready with limits, while public launch remains blocked until license, Stripe, dependency, registry, and legal gates close.
scout launch-readiness
scout launch-readiness --json
scout launch-readiness --require-public
The default command should report `Private beta: ready_with_limits` when the repo evidence is intact.
`--require-public` is expected to fail until the public launch blockers are closed.
Hosted beta
Hosted beta is a convenience API with finite credits. It is not unlimited crawling, and local Scout remains the primary beta path. Hosted beta remains invite-only and metered while pricing is derived from unit economics. Final credits, retention, and overage rules are not approved yet.
export SCOUT_HOSTED_BASE_URL="https://scout.chowmes.com"
export SCOUT_HOSTED_API_KEY="paste-your-delivered-key"
curl "$SCOUT_HOSTED_BASE_URL/v1/hosted/me" \
-H "Authorization: Bearer $SCOUT_HOSTED_API_KEY"
curl -X POST "$SCOUT_HOSTED_BASE_URL/v1/hosted/scrape" \
-H "Authorization: Bearer $SCOUT_HOSTED_API_KEY" \
-H "Content-Type: application/json" \
-d '{"url":"https://example.com","formats":["markdown"],"timeout_ms":30000}'
`/v1/hosted/me` shows your plan, credit balance, and current limits. A basic hosted scrape consumes one standard credit.
Hosted endpoints use `Authorization: Bearer ...`. Do not put hosted keys in frontend code, public repos, screenshots, or shared logs.
High-level hosted runs write records and source evidence into hosted artifact storage with limited beta retention.