How to connect Instagram to Grok Bot

Grok Bot only attaches MCP servers that are reachable over the public internet, so a local stdio server is refused. You do not need one. Ask Grok Bot to run npx -y @useclawlink/cli login in its terminal instead. The CLI is a shell command, not an MCP server. If you would rather have native tools, add https://claw-link.dev/api/mcp as a custom connector at grok.com/connectors and approve the ClawLink sign-in.

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

Instagram 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 Instagram 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.

This is the most common file failure we see across every client, and on Grok Bot it usually fixes itself. Instagram publishing needs the actual bytes, and a path inside Grok Bot's cloud computer, which looks like /home/box/sand-data/agents/<id>/attachments/<hash>.png, is not something our server can open. What makes Grok Bot different from most agents here is that it has a shell. The failed call hands back a single-use curl command in its hint, and Grok Bot can run it without being asked: it uploads the bytes to https://claw-link.dev/api/files/ingest, takes the file object from the JSON response, and calls the tool again with the s3key unchanged. Measured on a real run, 2026-08-30: the upload failed at 17:13:03 and the retry succeeded at 17:13:17, fourteen seconds later, with no human involved. So if you saw this error scroll past and then the post appeared anyway, that is what happened. Two things stop the loop. The upload link is single-use and expires in 15 minutes, so an agent that pauses and retries much later needs a fresh one, which means running the failing tool again. And if Grok Bot reports the error and stops instead of retrying, tell it to follow the curl instruction in the hint. A public HTTPS URL that serves the raw bytes works too and skips the upload entirely.

Almost every guide for this error tells you to add instagram_basic and instagram_content_publish and check your app's permission tier. That advice is for people running their own Meta app, and it is the wrong diagnosis on a hosted connection. Two things actually cause it here. The first is a partial approval: Meta's consent screen lets you decline individual permissions, so a connection can read your profile while every publish attempt fails. Reconnect from the dashboard and accept the whole set. The second, and the more common one with an agent, is an invented id. Grok Bot that has not listed anything yet will guess an ig-user-id or a media id, and the guess is structurally valid enough to reach Meta, which answers with a permissions error rather than a not-found. That is why it reads as a scope problem when it is a bad argument. Have Grok Bot run instagram_get_user_info for the real account id, then instagram_get_ig_user_media for real media ids, and retry with those.

The URL reached our server but did not hand back a file. A 403 usually means the link is signed, private, or blocks server-side fetches, which is common with a share link that works in your own logged-in browser. An HTML page means the link points at a viewer or landing page rather than the file, which is what a normal Dropbox or Drive share URL is. Google Drive and Dropbox share links are resolved automatically, so paste those unchanged. For anything else, use a direct link that serves the raw bytes, such as an S3, R2, or CDN URL. If you cannot produce one, skip the URL route: have Grok Bot upload the file from its own shell with the curl command in the failure hint above.

The account is a personal one. Meta's Instagram Graph API only serves Professional accounts, so the OAuth connection completes, Grok Bot reports success, and then every media call returns nothing. Switch to a Business or Creator account in the Instagram app under Settings, then Account type and tools, and link it to a Facebook Page if you plan to publish. Reconnect from the dashboard afterwards so the grant picks up the new account type. Worth separating from advice you will find elsewhere: the steps about creating a Meta app, requesting permissions, or submitting for App Review do not apply to this connection, but the Professional-account and linked-Page requirements do, because those are properties of your account rather than of the app.

Instagram publishing is two calls, not one. instagram_post_ig_user_media creates a container holding the image or video and the caption; instagram_post_ig_user_media_publish is what actually posts it. An agent that runs the first and stops has done half the job, and the post exists only as an unpublished container. Ask for the publish step explicitly. If the publish call itself fails, check the allowance with instagram_get_ig_user_content_publishing_limit, because a full window fails at exactly this step. Video adds a wrinkle: Meta processes it asynchronously, so a container can exist and not be ready to publish for a short period, and an agent that publishes immediately will see it fail.

This is an Instagram checkpoint on the account, not an expired ClawLink connection. Meta has flagged something and is holding the account until a human clears it. Open instagram.com in a normal browser, sign in, and complete whatever it asks. Reconnecting in ClawLink before you do that will not help, because the block is on the account rather than on the grant. Once the checkpoint is cleared the existing connection resumes working.

Try reconnecting Instagram 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 Grok Bot to Instagram: link your Instagram account once in the browser and Grok Bot can call the Instagram API through 29 ready-made tools — no custom code or token handling.

Ask Grok Bot to run npx -y @useclawlink/cli login in its terminal and approve the link it prints, then npx -y @useclawlink/cli connect instagram to authorize Instagram. Grok Bot calls the tools from the next message, and the pairing persists in its sandbox, so this is a one-time step: no config files or API keys to manage.

About two minutes. Sign in, click Connect next to Instagram in the dashboard, authenticate, and Grok Bot 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 Grok Bot. That is fine if you want to build and maintain the integration yourself. Most teams just want Instagram working from chat.

You can, and Grok Bot is capable of it, but understand what you are choosing. Browser login means your Instagram password and your 2FA prompt go through the agent's cloud session, the access is as broad as your own account, and there is nothing to revoke afterwards except changing your password. It also breaks whenever Instagram shows a checkpoint. A ClawLink connection is scoped API access instead: you approve it on Meta's own consent screen, Grok Bot never sees a password, and you can disconnect it from the dashboard. The browser route is genuinely better for the things the API does not expose, such as reading your own feed or interacting with Stories, so it is a reasonable fallback for those. For posting, comments, insights, and messaging, the API path is the one to use.

No. Doing this yourself means registering a Meta developer app, adding the Instagram Graph API, requesting instagram_basic and instagram_content_publish, and submitting for App Review before anyone other than you can use it. ClawLink connects through Meta's consent screen using an app that has already been through that process, so you sign in, approve, and the tools work. What you do still need is an Instagram Professional account, Business or Creator, linked to a Facebook Page if you intend to publish. That requirement belongs to your account rather than to the app, so no hosted provider can remove it.

Yes, and Grok Bot handles the hard part on its own. A file it just created lives inside its own cloud computer, which our server cannot read, so the first publish attempt fails with a missing-bytes error. Grok Bot then runs the single-use curl upload the error hands back and retries. On a measured run this took fourteen seconds start to finish and needed no input, which is why the usual experience is that it simply works. This is the one place where running in a sandbox with a terminal is an advantage: agents that can only read files from a fixed media folder have no way to recover from the same error. If you would rather avoid the round trip, put the file somewhere that serves the raw bytes over HTTPS and pass that URL instead.

Yes. Two separate things persist. The ClawLink pairing is stored in the sandbox at ~/.clawlink/credentials.json, so Grok Bot stays signed in to ClawLink between sessions. The Instagram grant itself lives on our side against your ClawLink account, not in the sandbox at all, so it is unaffected by anything that happens to the machine. If the sandbox is ever rebuilt from scratch you would re-run the login command; you would not reconnect Instagram.

Grok Bot only attaches MCP servers that are reachable over the public internet, so a local stdio server is refused. You do not need one. Ask Grok Bot to run npx -y @useclawlink/cli login in its terminal instead. The CLI is a shell command, not an MCP server. If you would rather have native tools, add https://claw-link.dev/api/mcp as a custom connector at grok.com/connectors and approve the ClawLink sign-in.