Call page.list_files

Answer a member's question about which FILES a site is built from — 'what files does this site have?', 'show me the code', 'which file is the hero in?', 'where does the pricing copy live?'. No site id is needed — it works on the site the member has open on screen. It is for showing and explaining the source; a change the member asks for goes to page.edit, whose coding agent finds the place itself. It is the same tree the Files board shows, for every site whose files we hold, however it was made or brought in. A site we hold no source for answers kind: no_source_snapshot; say so rather than inventing files. A BIG PROJECT ANSWERS IN PAGES: total is the real file count, and when more files follow the result carries next_cursor — call again with cursor: next_cursor and the same path_prefix and contains to read on (if the site's files changed in between, the cursor is refused: start again from the first page and read it over rather than joining the lists), or pass path_prefix (app/, components/, src/pages/) to walk one directory at a time. NEVER guess a path: if you have not seen it listed here, page or narrow until you have. CHECK can_write: when it is false this site's files are READ-ONLY here — answer questions about them and offer no edit, because every edit will refuse it. To answer WHERE some words live — the date, headline or sentence a member asks about — pass contains with a short fragment of them: only the files holding it are listed, each with match_line and match_text. Not for what is inside one file, or for making a change — that is page.read_file or page.edit.

Every result names the workspace it ran in. Say which workspace the answer is about. When the account has more than one workspace and none is selected, this tool refuses with workspace_ambiguous and lists the choices — offer them, never pick one.