openai/sharepoint-site-discovery

SharePoint Site Discovery

Use this skill when the main job is locating the right SharePoint site, drive, or folder before file analysis or editing.

Start Here

• Treat SharePoint discovery as site-scoped, not user-recency-scoped.
• Use search(query="...") for keyword search.
• Use search(query=None, hostname=..., site_path=..., folder_path=...) for browse mode.
• Use get_site(...) and list_site_drives(...) when the user knows the site but not the right library.

Workflow

1. If the user names a SharePoint hostname and site path, validate them with get_site(...).
2. If the site is known but the right library is not, use list_site_drives(...) to inspect the site-scoped document libraries.
3. If the user wants to browse a known folder or library, use search(query=None, hostname=..., site_path=..., folder_path=...) and inspect the immediate children.
4. If the user wants to find a file by keyword, use search(query="..."), then narrow with hostname, site_path, or folder_path when the scope is known.
5. Preserve the exact returned url, site, drive, and folder context so later fetch, update_file, or upload_file calls use the resolved destination instead of a guessed path.
6. When multiple plausible sites or libraries exist, present the candidates and explain the distinguishing context instead of picking silently.

Output Conventions

• Name the exact site, library, and folder you resolved.
• Distinguish clearly between browse results and keyword-search results.
• When handing off to another SharePoint workflow, include the resolved url or the exact site and folder context that should be reused.

Example Requests

• "Find the right SharePoint site for the launch checklist and show me the available document libraries."
• "Browse the ops site and narrow me to the folder that contains the Q2 roadmap files."
• "Search SharePoint for the pricing workbook, but keep the search inside the finance site."