How to connect YouTube to Hermes Agent

Pairing is a two-step handshake: run hermes clawlink begin, approve the link in your browser, then run hermes clawlink finish. If you ran finish before approving, or the approval link expired, run hermes clawlink begin again to get a fresh link. Confirm the plugin was installed with --enable, then verify with hermes clawlink test.

Reconnect YouTube from the dashboard, then start a fresh chat if the runtime still has the old tool catalog loaded.

YouTube tool schemas load on demand the first time a tool runs and are cached after that, so this error usually clears on its own: wait a few seconds and retry the same request. If every YouTube call keeps failing with it in a fresh chat, reconnect from the dashboard, and contact support if it still persists — that pattern points to a configuration problem on our side, not something you can fix by reconnecting again.

Two usual causes. The connected account may not have access to the specific workspace, inbox, store, or project in the request — check that first. If access looks right, the agent may have sent a placeholder value (like "YOUR_ID" or an example id from documentation) instead of a real one: ask it to run a list or search tool first, then retry the action with a real id from those results. Most failures at this stage are one of these two, not ClawLink bugs.

On a self-built integration this means your OAuth client did not request the right scope, and the fix is a console trip. On a hosted connection it usually means something else. The first cause is a partial approval: Google's consent screen lets you untick individual permissions, and unticking one leaves the connection active but unable to do that class of action, so reads work and writes fail. Reconnect and accept the full set. The second is channel scope, which is the Brand Account case: the grant is attached to the channel you chose at sign-in, so acting on a different channel returns a permissions error rather than an empty result. Ask the agent to run youtube_get_channel_statistics and read back which channel it sees before you change anything else.

Ask the agent to diagnose it:

Run youtube_get_channel_statistics and tell me the exact channel title and id you are connected to. Then run youtube_list_channel_videos and report how many videos you can see. Do not retry the failed call yet.

Reconnecting does nothing for this one, and neither does waiting a few minutes. Google's quota resets at midnight Pacific Time, and per its current documentation a project gets 100 search.list calls, 100 videos.insert calls, and 10,000 units a day for the remaining endpoints. Searching burns its own counter fastest: an agent asked an open question will often search several times to refine, and 100 searches disappear quickly. The fix is to change the shape of the request rather than to retry it. Search once, keep the video ids, then use youtube_get_video_details_batch, which pulls many videos in one call, and youtube_list_channel_videos, which enumerates a channel without touching the search allocation at all. Note that Google's older published quota table, which charged 100 units per search and 1,600 per upload, is still what most guides and AI answers repeat; the numbers here are from Google's current getting-started page.

Ask the agent to diagnose it:

Tell me exactly which YouTube tools you called in the last few minutes and how many times each. Do not call any more YouTube tools until I reply.

Tool schemas load on demand, so the first YouTube call in a new session can arrive before the catalog does. Retry once and it normally resolves. If the same name fails a second time, the name is wrong rather than missing, and the error lists the closest real tools. The doubled prefix shape is the common invention: the tools are named youtube_search_you_tube and youtube_list_channel_videos, with the integration prefix already included, so an agent adding another one produces a name that has never existed. This is a different failure from the one most search results describe, which is a local MCP server missing from claude_desktop_config.json; there is no local server here and no config file to edit.

Ask the agent to diagnose it:

List the YouTube tools you actually have access to. Then tell me which one searches YouTube and use that exact name.

Try reconnecting YouTube and complete the consent flow in the same browser session. Partial OAuth approvals or switching accounts mid-flow can leave the connection incomplete.

Yes. ClawLink is the fastest way to connect Hermes to YouTube: link your YouTube account once in the browser and Hermes Agent can call the YouTube API through 47 ready-made tools — no custom code or token handling.

Install the plugin with hermes plugins install ClawLink-HQ/hermes-plugin --enable, then pair once: run hermes clawlink begin, approve the link in your browser, and run hermes clawlink finish. Connect YouTube in the dashboard and Hermes can use it from the next message — no config files or API keys to manage.

About two minutes. Sign in, click Connect next to YouTube in the dashboard, authenticate, and Hermes Agent can use it from the next chat message.

The alternative to ClawLink is usually manual OAuth app setup plus your own token handling, permission troubleshooting, and tool plumbing for Hermes Agent. That is fine if you want to build and maintain the integration yourself. Most teams just want YouTube working from chat.

No. That requirement comes from building the integration yourself: enable YouTube Data API v3 in a project you own, configure an OAuth consent screen, and handle the verification process before anyone outside your test list can use it. ClawLink's connection runs through an already-registered app, so the only thing you do is approve access to your channel on Google's own consent screen. This also removes the trap that catches most DIY attempts, where an app left in testing mode issues refresh tokens that stop working after seven days.

Google allocates quota per project, and per Google's current documentation a project that enables the YouTube Data API gets 100 search.list calls, 100 videos.insert calls, and 10,000 units a day for everything else. Because the connection runs through a shared app rather than a project you own, you are not managing that allocation and you cannot raise it from your side. In practice the limit people meet first is the search call count, not the unit pool, since searching is capped at 100 calls a day on its own counter. If your agent does heavy discovery work, ask it to search once and then work from the returned video ids with youtube_get_video_details_batch instead of searching repeatedly.

Yes, through youtube_upload_video or youtube_multipart_upload_video, and this is the one action worth being deliberate about. An upload writes to your public channel, so decide up front whether you want the agent to publish or to leave the video private for review, and say which in the prompt. Uploading also runs against the videos.insert call allocation, which Google documents as 100 calls a day.

Yes, but you have to pick the channel during the Google sign-in rather than after. A Google login that manages several channels will present a channel chooser, and the grant attaches to whichever one you select. If the agent later reports that your videos do not exist, it is almost always connected to your personal channel instead of the Brand Account. Reconnect from the dashboard and choose the right channel on that screen.

Pairing is a two-step handshake: run hermes clawlink begin, approve the link in your browser, then run hermes clawlink finish. If you ran finish before approving, or the approval link expired, run hermes clawlink begin again to get a fresh link. Confirm the plugin was installed with --enable, then verify with hermes clawlink test.