FeynmanLM Docs

Appearance

Connectors ​

Connectors are optional. Open Settings -> Connectors to see each connector and its connection status, then select one to set it up.

Connectors only run when you explicitly use them. FeynmanLM never scans X, Google Drive, YouTube, or Read AI on app startup — automatic imports, where offered, are opt-in and run in the background after launch.

X Bookmarks ​

Use the X connector to sync saved X bookmarks into Studio as sources.

Reading bookmarks requires a paid X API plan. X does not include bookmark access on the free tier.

Set Up X ​

The one-time X Developer Portal setup is fiddly, so have your AI assistant do it. Paste this into an AI agent that can browse the web (Claude, ChatGPT with browser access, or similar):

text
Set up X (Twitter) API access for me so the FeynmanLM app can sync my X
bookmarks. Drive the browser yourself where you can; hand back to me only
for logins, 2FA, and payment confirmation.

1. Go to https://developer.x.com/ and sign in with my X account (I'll take
   over for the login if needed).
2. If I don't have a developer account yet, complete the signup first:
   accept the developer agreement, and if asked for a use-case
   description, say I'm syncing my own bookmarks into a personal
   learning app (read-only, personal use). X may require a verified
   email or phone number on my X account — hand back to me if so.
3. Bookmark access requires a paid X API plan. Check my current plan; if
   I'm on the free tier, look up what the Basic plan (or X's current
   pay-as-you-go option) costs right now, tell me the price, and take me
   to the subscription page so I can confirm the purchase myself — do
   not pay on my behalf.
4. Create a Project and an App inside it (any name is fine, e.g.
   "FeynmanLM"). If the plan signup already created a default Project
   and App, reuse those instead.
5. In the App's settings, open "User authentication settings" and set:
   - App permissions: Read
   - Type of App: Native App (public client)
   - Callback URI / Redirect URL: feynmanlm://x-callback
     (exactly this — anything else makes X reject the sign-in)
   - Website URL: any valid URL I own, or my X profile URL
   Then save.
6. Open the App's "Keys and tokens" tab and copy the OAuth 2.0 Client ID —
   not the API Key, API Key Secret, or Bearer Token.
7. Give me that Client ID and remind me to paste it into FeynmanLM under
   Settings -> Connectors -> X, then click Connect.

When your assistant hands back the Client ID:

  1. In FeynmanLM, open Settings -> Connectors and select X.
  2. Paste the Client ID and click Connect.

After connecting, use Sync on the X connector page or Sync Bookmarks in the Studio X tab.

Paid plan required

The free X API tier does not include bookmark access.

Prefer to do the portal setup by hand?
  1. Open the X Developer Portal and sign in.
  2. Subscribe to the Basic plan or higher.
  3. Create an app under Projects & Apps.
  4. Under User Authentication, configure:
    • App permissions: Read
    • Type of App: Native App
    • Callback URL: feynmanlm://x-callback
    • Website URL: your website or another valid URL required by X
  5. Copy the app's Client ID from the Keys and Tokens tab.

Google Drive ​

Use the Google Drive connector to import Drive folders as meeting sources — folders containing Google Meet transcripts, notes, chats, recordings, and attachments.

Google Drive access is read-only and scoped to exactly what you pick. FeynmanLM uses Google's own file picker, so it can only ever see the folders and files you select — nothing else in your Drive.

Set Up Google Drive ​

There is no Google Cloud setup. FeynmanLM signs in against its own OAuth client:

  1. In FeynmanLM, open Settings -> Connectors and select Google Drive.
  2. Click Connect and sign in with your Google account.

FeynmanLM opens a temporary 127.0.0.1 callback while Google sign-in is in progress, and requests this Google scope:

text
https://www.googleapis.com/auth/drive.file

drive.file is a non-sensitive scope: it grants access only to files you explicitly open with FeynmanLM through the picker, so there is no whole-Drive access and no unverified-app warning.

Import a Folder ​

  1. Open Studio -> Meetings.
  2. Click Import Meeting Folder....
  3. Enter a group name, then click Choose Folder in Google Drive….
  4. Pick the meeting folder in Google's file picker. FeynmanLM imports the transcript, notes, chat, and recording files it finds inside.

A public Drive folder URL still works without connecting (paste it and choose Import Drive URL), and the local folder import path still works if you prefer Google Drive for desktop or a downloaded folder.

External assistants (MCP)

Because drive.file only exposes files you pick in the app, an AI assistant connected over MCP can't reach an arbitrary private Drive file by URL — select it in FeynmanLM's picker first. Refreshing already-imported Drive sources over MCP keeps working. The search_google_docs tool has been removed, since whole-Drive search isn't possible under drive.file.

YouTube ​

Use the YouTube connector to follow your Liked videos. Liking a video on YouTube is how videos enter Studio: new likes are imported automatically in the background, so you don't have to add each video by hand.

The connector imports metadata only — a video's transcript (and slides) are fetched on demand the first time you open or review it, not when it's imported.

Connecting YouTube is read-only and is used only to read your Liked videos feed, which is private to your account.

Set Up YouTube ​

  1. In FeynmanLM, open Settings -> Connectors.
  2. Select YouTube and click Connect. Google asks you to grant read-only access to your YouTube account.

There is no Google Cloud setup: YouTube signs in against FeynmanLM's own OAuth client, just like the Google Drive connector above.

FeynmanLM opens a temporary 127.0.0.1 callback while Google sign-in is in progress, and requests this Google scope:

text
https://www.googleapis.com/auth/youtube.readonly

Follow Your Liked Videos ​

  1. After connecting, the YouTube connector page shows a Liked videos row.
  2. Click Follow. Your current liked videos import right away, and Liked videos appears under Following.
  3. To stop importing new likes without deleting what's already imported, click Unfollow.

Channels and playlists cannot be followed — like a video on YouTube to bring it into Studio. (Playlist URLs added from the Courses tab still import as course lectures.)

Automatic Import ​

Turn on Automatically import new videos on the YouTube connector page to keep your Liked videos fresh. FeynmanLM re-checks them shortly after launch and every few hours, importing only new videos. It is off until you follow Liked videos, and it never runs at app startup.

Watch Later is not available

Google's YouTube Data API does not expose the Watch Later playlist, so it can't be followed. Your Liked videos feed is fully supported.

Read AI ​

Use the Read AI connector to import your Read AI meeting transcripts as meeting sources in Studio — each meeting becomes a session under its recurring meeting group.

Set Up Read AI ​

Read AI has no static API keys yet, so connecting works through a browser sign-in (OAuth 2.1):

  1. In FeynmanLM, open Settings -> Connectors and select Read AI.
  2. Click Connect. FeynmanLM registers a private OAuth client and opens the Read AI sign-in page in your browser.
  3. Sign in and approve access.

FeynmanLM stores only the resulting tokens in your Keychain and uses them read-only, to list meetings and fetch transcripts.

Workspace setting required

Your Read AI workspace must have Downloads enabled under Workspace Settings → Reports & Sharing. Without it, transcript fetches fail.

Fetch Meetings ​

  1. Open the Read AI tab in Studio.
  2. Click Add Meetings… -> Fetch from Read AI….
  3. Select the completed meetings you want, choose a meeting group, and click Import.

Already-imported meetings are skipped, and meetings whose transcript isn't ready yet are reported rather than guessed.

Automatic Import ​

Turn on automatic import on the Read AI connector page to have FeynmanLM check for new finished meetings shortly after launch and every few hours, importing new transcripts in the background. It is off by default and never runs at app startup.

Import Without Connecting ​

If you'd rather not connect the API, export a transcript from the Read AI dashboard and use Add Meetings… -> Import Read AI Transcript File… in the Studio Read AI tab.

External assistants (MCP)

Once connected, an AI assistant can list and import your meetings with the list_readai_meetings and import_readai_meetings MCP tools — see the MCP Tools Reference.