What problem does it solve? Research sessions lose track of where web evidence came from, and uncontrolled crawling risks touching private or unauthorized content. This Skill acquires public web sources under explicit domain, page, byte, and time limits, then records provenance so later sessions can audit exactly what was fetched and how. ## Core Features & Use Cases - Bounded public acquisition: Fetch, search, and render public pages with enforced allowlists, redirect checks, and content-hash provenance records (ds-lite.source-record.v2). - Backend capability discovery: Detect and record available backends (Playwright CLI, Firecrawl, Tapestry, agent-browser) as ds-lite.capability.v1 instead of guessing or auto-installing. - Strict safety policy: Blocks login state, cookies, form submission, and unrestricted crawling; Firecrawl requires both an API key and explicit authorization. - Use Case: While surveying baselines for a paper, capture a set of public project pages and PDFs within a defined domain scope, producing auditable source records for later citation verification. ## Quick Start Ask the agent to use ds-lite-web to fetch public pages from an allowed domain list with page and byte limits, then report the recorded source provenance.
