Terms and Privacy

What Google's terms say about this route, as read on 2026-10-15, and what nobody has tested. This page does not call the route legal. It says what was found and where.

Google Search is out

The Gemini API Additional Terms (effective 2026-03-23) restrict "Grounded Results", the answers Gemini gives with its Google Search tool: no caching, no analysis or mixing by others. They also forbid using that search "to identify destination pages for crawling or scraping". So fetchwright never sends the Google Search tool and never uses Gemini to find URLs. Find them some other way and hand them in.

URL context

No clause in the Additional Terms names URL context, and the URL context docs list technical limits only. Whether URL context output counts as a "Grounded Result" is not stated anywhere. The general rule applies: "You're responsible for your use of generated content, and for the use of that content by anyone you share it with."

The site's side

Untested. For Reddit, Google reads its own copy of the page and hands back its reading. No clause was found that forbids reading Google's copy this way, and none that permits it.

What you send, and who may see it

fetchwright sends Google a public URL and an instruction, and gets the page's text back. Nothing else.

• On the free tier, Google may use what you send and receive to improve its products, and human reviewers may read it. Google's terms say not to send sensitive, confidential or personal information. Read public pages only.
• On a paid project (a billing account linked to it), prompts and answers are not used that way. Google logs them for a limited time to detect abuse.
• Requests carry store: false, so the Interactions API does not keep the exchange.
• You must be 18 or older to use the Gemini API.
• A tool that serves users in the EEA, Switzerland or the UK may use only paid Gemini services.

What fetchwright itself keeps

Nothing. It writes no logs and no cache. It writes files only for --out and --save-raw, and only where you say. --save-raw files hold page text, so keep them out of repositories. The key is read from the environment and never printed or written.

If you write notes from what it reads

The skill asks the agent to paraphrase, quote at most a line, link the source, keep usernames out, and leave out what an author deleted. fetchwright already drops placeholders and lone usernames (How pages are read).

Sources: ai.google.dev/gemini-api/terms (effective 2026-03-23), ai.google.dev/gemini-api/docs/url-context (updated 2026-10-13), ai.google.dev/gemini-api/docs/interactions (updated 2026-10-15). The fetchwright-setup skill re-checks these before quoting them.