Just Scrape Skill Security Audit

If the CLI is absent, the Skill globally installs `just-scrape@latest` and immediately runs it. `latest` changes over time, and a global install modifies the user's command-line environment. The supplied source does not include the npm package implementation, so its install scripts and runtime behavior cannot be audited here. The user can require a pinned, verified version installed in an isolated, non-global environment.

The examples explicitly pass a session cookie and bearer token to a CLI used for ScrapeGraph AI requests to obtain account statistics or protected tables. This lets the CLI and its remote processing chain use the credentials to access authenticated pages. Environment variables avoid literal secrets in the command but do not prevent the CLI from receiving them. The user should confirm whether the service forwards, stores, or logs these values and use short-lived, least-privilege credentials only for authorized data.

By default, the Skill writes search, page, and structured extraction results into the project's `.just-scrape/` directory and can retrieve detailed JSON for earlier requests. If scraped content or historical payloads contain personal, account, or internal data, persistent local copies may remain. `.gitignore` only reduces Git-commit risk; it does not encrypt, restrict access, or clean files. The user can require in-context output only, a protected temporary directory, and deletion of sensitive results after use.

The Skill directs the agent to make a small real request before actual work, while stating that every operation consumes credits, so even setup verification may incur a charge. The three parallel scrapes are an example rather than a mandatory step, and it checks credits first and warns about expensive options; executing it would nevertheless initiate several billable operations at once. The user can require skipping paid test calls, set request/page limits, and approve a budget before each batch.

`monitor create` establishes recurring hourly or cron-based scraping and can send notifications to a webhook supplied through an environment variable. Because ScrapeGraph operations consume credits, an active or resumed monitor can continue generating charges; if the webhook belongs to an external recipient, change notifications also leave ScrapeGraph. Pause and delete commands exist, but creation does not require an expiry, spending cap, or recipient confirmation. The user can require those limits and an approved webhook before creation.