FAQ The Quiet Questions

Questions,
answered.

The stuff people wonder about but don't always ask out loud. If yours isn't here, send it in and we'll add it.

No. Claude Code only sees the folder you open it in (plus anything you explicitly point it at, like a connector). It has no ambient access to other repos, other studios' code, or anything outside the session you're in.

Under Anthropic's standard commercial terms, conversations and code from paid accounts like our studio's aren't used to train models. That's the general policy — if you want the specifics of our exact contract, ask whoever manages our Anthropic account, since terms can vary by plan.

The same thing that happens when any teammate gets something wrong: it goes through review before it ships. Claude's output is a strong draft, not a final answer — code goes through PR review, content goes through the same checks anyone's work would. If it's confidently wrong in a way that would fool a reviewer, that's worth flagging as a skill gap, not just fixing quietly.

Not for studio code, projects, or anything confidential — use the Lasting Light account your lead set up. Personal accounts don't carry our team's skills, MCP connections, or data-handling terms. It's fine to use a personal account for your own unrelated stuff, obviously — just not ours.

This site exists to make the job better, not to shrink the team — the whole framing is "Claude handles the pipeline scaffolding so you spend more time on the craft that's actually yours." Headcount decisions aren't something this page can speak to; if that's a real concern, that's a conversation for leadership, not something to infer from a training site.

Yes — see Everyday Claude. Writing, meetings, spreadsheets, research, and learning make up most of most people's week, and none of it requires touching Unreal or a terminal.

No. Claude can open PRs and stage changes, but merging, deploying, and shipping to players or partners all still go through a human — same review gates as anyone's work, no exceptions.

A skill is a written playbook — our conventions, commands, and checklists, encoded as text Claude reads. A connector is a live wire to another tool (Slack, Drive, our Unreal editor) that gives Claude new abilities. Skills tell it how we do things; connectors give it something to do them with.

The Contribute page's form covers site problems too — pick "Something's broken on this site" from the dropdown. Thirty seconds, no login needed.

Not the suggestion form — go straight to your lead or however the studio already handles security reports. If credentials or secrets were exposed anywhere (including inside a chat with Claude), say so immediately rather than trying to quietly fix it yourself.

← Prev
Week Two