Cosdesign Standalone

Plain-language considerations before installing or running this skill: - What it actually does: the packaged JS files only print structured prompts; the actual webpage fetching and extraction will be performed by your agent/platform (WebFetch or Jina Reader). The skill delegates network access to the platform tools, it does not fetch sites itself. - Privacy risk: when you ask it to analyze a URL, the full page content (HTML/CSS/computed styles) will be sent to the model/tooling and possibly to third-party fetch services (r.jina.ai). Do not use this on private, behind-login, or sensitive pages unless you understand where that data will be sent and stored. - Inconsistencies to note: the README claims 'no downloading of fonts/images' but the export/html-report template asks to 'use the site's actual fonts' (which may imply fetching fonts). Also the README claims analysis history is appended to data/analysis-history.json, but the provided scripts do not write to that file — so persistence behavior is unclear. - Credentials: JINA_API_KEY is optional. Only provide it if you trust Jina Reader and need higher crawl throughput. - Mitigations: review or run the scripts locally in a sandbox; avoid feeding private/authenticated URLs; inspect any output before sharing it externally; if you don't want any persistence, remove or clear data/analysis-history.json and run the scripts manually to confirm they do not write files; if you require stricter guarantees, ask the author for a version that explicitly avoids any external fetch services or that documents exactly which services are contacted and what data is transmitted.