The deep view
On schema, rate, and the ethical posture.
The point of the Web Scraper is the schema. The agent describes the target data as a typed schema — fields, types, validation. The scraper writes the data as the schema. The diff is reviewable. When the target page changes, the reviewer agent flags the schema drift before the data is allowed to commit. The override is logged. The audit trail is the scraper log.
Rate-awareness is a default, not a setting. robots.txt is respected. The scraper throttles per host, retries on 429, surfaces the rate-limit log to the operator. The agent does not exhaust the host. The override is logged. The reviewer agent reads the rate-limit posture before the scrape runs. The policy is explicit.
The ethical posture is the part other scrapers skip. The Web Scraper respects the site's terms. It does not bypass CAPTCHAs. It does not scrape behind a login without explicit operator consent. The reviewer agent reads the policy before the scrape runs. The override is logged. The audit trail stays honest.
The MCP server is the integration story. Any agent in the Netwit runtime can request a scrape via the typed scraper MCP server. The contract is the contract. The contract drift is caught by the reviewer agent before the call is allowed. The scraper log lands in the repo — every request, every response, every rate-limit event, every schema change. The operator can read it, version it, branch it, hand it to another team.