# Ingesting from Dropbox RAGfly can read documents **straight from Dropbox's API** — no local sync client, no downloads, no copying folders to disk. This guide is for the **workspace administrator** who enables the connector once. After that, every user in the group just picks a folder and RAGfly indexes it. > **One-time setup, ~5 minutes.** You do it once per RAGfly group. Your users > never see any of this. --- ## What you get - Files are read **directly from Dropbox's API**, including files that live only in the cloud and were never downloaded to a computer. - A built-in **folder browser**: users navigate their Dropbox and pick the folder to index (Dropbox has no folder picker of its own, so RAGfly provides one). ## Privacy: the file never leaves your browser This is the same guarantee as loading from a local folder or from Google Drive, and it is not negotiable: 1. Bytes are downloaded from Dropbox **into your browser's memory (RAM)**. 2. Text extraction runs **in your browser**. 3. Only **encrypted text** is uploaded to RAGfly. The original file is never stored by RAGfly and never travels to RAGfly Cloud. RAGfly requests **read-only** permissions (`files.metadata.read` and `files.content.read`): it can read, never modify or delete. --- ## Before you start You need: - A **Dropbox** account — https://www.dropbox.com/developers/apps - **Administrator** access to your RAGfly group (to edit Group Parameters) - The **domain** where you use RAGfly: `https://app.ragfly.ai` (or your own) --- ## Part 1 — Create the app in Dropbox Unlike Google Drive, Dropbox needs **one credential only**: the **App key**. RAGfly uses OAuth 2.0 with PKCE, a flow designed for browser apps, so the **App secret is never used** — don't copy it anywhere. ### 1.1 Create the app Go to https://www.dropbox.com/developers/apps and click **Create app**: - **API**: **Scoped access** - **Type of access**: **Full Dropbox** (so users can pick any folder of their account; "App folder" would confine them to a single sandbox folder) - **Name**: anything (e.g. "RAGfly Connector") ### 1.2 Grant the read-only permissions In the app's **Permissions** tab, check: - `files.metadata.read` — list folders and files - `files.content.read` — download file contents Click **Submit**. Nothing else — RAGfly never asks for write permissions. > ⚠️ Permissions are baked into each token when the user authorizes. If you > change them later, users must re-authorize (use **Switch account**). ### 1.3 Register the redirect URI In the app's **Settings** tab, under **OAuth 2 → Redirect URIs**, add: ``` https://app.ragfly.ai/dropbox-callback ``` One line per environment where you use RAGfly, always ending in `/dropbox-callback`. > ⚠️ Miss this step and users get a **redirect URI mismatch** error when they > try to authorize. ### 1.4 Raise the user limit Still in **Settings**, look for **Development users** — a new app says "Only you", meaning nobody else can authorize it. Click **Enable additional users** to raise the cap to 500. Skip this and every colleague gets *"This app has reached its user limit"* the moment they try to connect. ### 1.5 Copy the App key Still in **Settings**: copy the **App key** (a short lowercase string). Ignore the App secret — the PKCE flow doesn't use it. --- ## Part 2 — Enable the connector in RAGfly Sign in to RAGfly as a group administrator and go to **Group Parameters**. ### 2.1 Load the credential Under **Ingestion connectors**, paste it: | Parameter | Value | |---|---| | `Dropbox — App key` | the App key from step 1.5 | It is stored **per group**: each customer uses their own Dropbox app, and the value is shown masked. ### 2.2 Turn the connector on Under **Feature flags**, switch **`Dropbox connector`** to `true`. That's it. The connector appears **only** when both conditions are met: the flag is on **and** the App key is loaded. --- ## Part 3 — Using it (this is what your users do) In **Documents → Feed documents**: 1. In **Select source**, choose **Dropbox**. A status badge shows whether you are connected. 2. Click **Connect folder…**. Dropbox asks you to authorize (read-only), then RAGfly opens its folder browser showing **your Dropbox in the cloud**. 3. Navigate and click **Use this folder** on the folder you want. The badge turns to **Connected**. 4. Continue as usual: **Load directories**, then **Vectorize**. **Switching Dropbox accounts**: use **Switch account**. This forgets the current authorization and lets you pick a different account — the linked folder is cleared, since it belonged to the previous account. ### The status badge | Badge | Meaning | |---|---| | **Disconnected** | No folder linked yet. Click "Connect folder…". | | **Needs authorization** | A folder is linked, but the session expired (this happens after reloading the page). You will be prompted when you run. | | **Connected** | Folder linked and authorization live. | --- ## Troubleshooting ### Redirect URI mismatch when authorizing The domain you are using is not listed in the app's **Redirect URIs** (step 1.3). Add it exactly, with `https://` and the `/dropbox-callback` path. ### "This app has reached its user limit" A brand-new Dropbox app is limited to **one** user — its own developer — so the first colleague who tries to authorize it hits this error. In the app's **Settings** tab, find **Development users** ("Only you") and click **Enable additional users**: the cap rises to 500, which covers any pilot. Beyond 500 users you need **Apply for production**, a formal Dropbox review that also requires the branding of step 2 to be complete. ### The browser blocked the authorization popup Allow popups for your RAGfly domain and click **Connect folder…** again. ### The connector doesn't appear at all in the source selector Both conditions must hold, in Group Parameters: - the flag `Dropbox connector` is `true`, and - the `Dropbox — App key` credential is loaded. If you just changed either one, reload the page. ### "Missing scope" or permission errors when listing folders The app was authorized before you granted the permissions of step 1.2. Grant them, then use **Switch account** to re-authorize. ### Some files are skipped RAGfly ingests documents it can extract text from. Files with no extractable text (images without OCR, unsupported binaries) are reported as such in the pipeline, not silently dropped. --- ## Frequently asked **Does RAGfly store my files?** No. Bytes are read into your browser, text is extracted there, and only encrypted text is uploaded. The original file never reaches RAGfly Cloud. **Can RAGfly modify or delete my Dropbox?** No. The requested permissions are read-only (`files.metadata.read`, `files.content.read`). **Why isn't the App secret needed? Is the App key safe to store?** RAGfly uses OAuth PKCE, the flow made for browser apps: the secret is never involved. The App key is a public identifier by design — the real secret is the access token each user grants in the popup, which lives only in the browser's memory and is never persisted. Abuse is prevented by the redirect URI allowlist, not by hiding the key. **Do I need Dropbox installed on my computer?** No. Everything goes through Dropbox's API. Files that only exist in the cloud work fine. **Can each user connect their own Dropbox?** Yes. The App key is per group (the app), but each user authorizes their own Dropbox account and picks their own folder. **Do I have to repeat this per environment?** The credential is per group. If you use more than one RAGfly environment, add each `/dropbox-callback` redirect URI in step 1.3.