Integrate MediaCrawler

Default to a dry-run integration plan. The upstream license permits learning/research, not commercial use.

Workflow

Read references/integration.md before any installation or command proposal.
Confirm the purpose is non-commercial research and the targets/data are owned, public with permitted collection, or explicitly authorized. Otherwise stop.
Resolve the repository and verify remote, pinned commit, license, and worktree.
Produce a bounded run manifest: platform, login mode, crawl type, exact IDs/keywords, maximum items, comments off/on, concurrency, output path, retention, and deletion plan.
Default to no proxy, no browser-profile reuse, no comments, concurrency 1, minimum item count, and a new isolated output directory.
Show installation and run commands before execution. Treat QR/phone/Cookie login as a separate sensitive action.
Never print or persist Cookie values in chat, logs, plans, or receipts.
Execute only after the user approves the bounded manifest and any login/browser access.
Verify output count and destination, then stop; do not expand scope automatically.

Refuse commercial use, large-scale collection, private data, platform-control evasion, signature reverse engineering, CAPTCHA bypass, rotating proxies, or scraping that violates terms/robots/law.
Do not connect to an existing Chrome profile unless the user explicitly authorizes that exact profile and read scope.
Do not describe structural setup as runtime validation.

Return the authorization basis, repository pin, dry-run manifest, commands, records created, secrets handling, and compliance limits.