Google Drive
Ingest Google Docs, Sheets, Slides, and PDFs into the workspace knowledge graph. File metadata and extracted text feed the LLM extractor so the topics, people, and concepts in your documents become typed nodes.
Google Drive
Docs, Sheets, Slides, PDFs → `document` nodes + extracted entities.
The Drive connector scans Google Workspace native files (Docs, Sheets, Slides) plus PDFs and plain-text files in the authenticated user's Drive. Each file becomes a document node with owner, timestamps, and a content-type-aware text export feeding the LLM extractor — so the entities mentioned in your documents land as typed person / organization / concept / task / date nodes wired back to the source document.
What gets ingested
| Source | Node type | Edge type | Direction |
|---|---|---|---|
| Drive file (Doc / Sheet / Slide / PDF / text) | document | — | — |
| File owner | person | owns / owned_by | Person ↔ Document |
| Entity extracted from exported text | varies | mentions | Document → Entity |
Real use cases
- Find the deck — semantic search across exported Slides text retrieves the right deck by topic, not just by filename.
- Document-aware briefings — "summarize every Drive document that mentions Acme" runs as one graph traversal followed by an LLM rollup against the embeddings.
- Source-grounded answers — when an agent answers a workspace question, the source
documentnodes are linked so the user can click through to the original Drive file.
Settings
| Key | Type | Default | Description |
|---|---|---|---|
backfill_days | number | 365 | How far back to scan. Subsequent syncs are incremental (only files modified since the last run). |
extract_content | boolean | true | When false, files are stored as metadata-only nodes — no LLM extraction. |
include_shared | boolean | true | When false, only files the user owns are scanned. |
max_file_bytes | number | 10,000,000 (10 MB) | Files whose exported text exceeds this size are stored as metadata-only nodes. |
mime_types | string[] | Docs, Sheets, Slides, PDF, plain text | Restrict to specific content types. |
folder_id | string | "" | When set, only files directly inside this folder are scanned. Use the folder picker in the connection drawer. |
OAuth scopes
drive.readonly— read file metadata and contentuserinfo.email— resolve the connecting user's email
Read-only — no upload, no edit, no delete.
Google Contacts
Import your Google Contacts as canonical `person` nodes in the workspace knowledge graph. Establishes the people corpus your other connectors dedupe against.
GitHub
Sync a GitHub repository into the workspace knowledge graph as a typed code-graph — files, functions, classes, imports, calls, commits — keeping the graph fresh on every push.