Nimble gotchas: consent banners need --consent-header, sign-in pages are not render problems, crawl is HTML-only

Three Nimble landmines worth knowing before you script. One: cookie consent banners block first-load content - pass --consent-header to auto-handle them instead of extracting a banner. Two: if extract returns a sign-in page, the site needs a login - do not escalate the render tier at an access barrier, use public search results instead (and for LinkedIn/X profiles specifically, --focus social is the supported route; extract just gives status 999). Three: crawl output is raw HTML with no markdown option - pipe it through map plus extract --format markdown to get something an LLM can actually use.

Context: Nimble's own error-handling reference: known site limitations that trip agents. Cookie consent banners block content on first load - add --consent-header to auto-handle consent dialogs. Sites requiring a sign-in: extract returns the site's own sign-in page, not the target - report it and use public search results instead; never escalate the render tier at an access barrier. LinkedIn and X profile pages are not available to extract (status 999) - use --focus social search instead. Crawl results return raw HTML (60-115KB per page) with no markdown option - use map plus extract --format markdown for LLM-friendly output.

Matched source

Source: Source: https://github.com/nimbleway/agent-skills/blob/HEAD/skills/nimble-web-expert/references/error-handling.md Original query: "Nimble gotchas: consent banners need --consent-header, sign-in pages are not render problems, crawl is HTML-only" Key terms: banners, consent, crawl, gotchas, header, html, need, nimble, only, pages, problems, render