How to connect LinkedIn 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 LinkedIn from the dashboard, then start a fresh chat if the runtime still has the old tool catalog loaded.

LinkedIn 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 LinkedIn 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.

Most guides for this error walk you through the Developer Portal: add scopes, request Products, clear a token cache, add a LinkedIn-Version header. None of that applies here, because you have no app and no token cache. What the engines are really teaching is scope-first thinking, and the scoped version of this error is real: posting as an organization fails because this connection does not hold w_organization_social, and commenting on a post from a company page your account has no role on fails the same way. The second cause is a stale grant: if the account's consent has been revoked or the connection expired, reads keep working through a cached session while writes fail. Reconnect from the dashboard and complete the consent screen again. Do not ask the agent to fabricate scopes or re-request the post, and note that a placeholder author URN produces this same shape and is covered in its own card below.

Ask the agent to diagnose it:

Run linkedin_get_my_info and report what it returns. Then retry the failing call once and quote the exact error body, including any status code.

The create-post tool requires an author argument in URN form, urn:li:person:\<id> for a person, and agents invent it or skip it when they have never listed anything. This is the most common failure on this integration in production. The fix is list-first: run linkedin_get_my_info to get your own profile, then call linkedin_create_linked_in_post with the author URN built from that profile's person identifier and your commentary. A placeholder string that resembles a URN but is not yours produces a different, later error, so the order matters.

Ask the agent to diagnose it:

Run linkedin_get_my_info and show me the profile response. Then retry the post with the author value set to the person URN from that response. Do not guess the author value.

The tool is named linkedin_create_linked_in_post. If the agent reached for a shorter name like linkedin_create_post or linkedin_post_share, the error message lists the real names, so retrying with one of those works. If the name you asked for already matches the list exactly, the tool catalog had not loaded when the first call went out; schemas load on demand, so one retry after a few seconds clears it. The same pattern shows up for reads: linkedin_get_profile and linkedin_get_me do not exist; the real tools are linkedin_get_my_info, linkedin_get_image, and linkedin_get_images.

Ask the agent to diagnose it:

List the LinkedIn tools you actually have access to, using the exact names. Use one of those names for what I asked.

Three distinct causes produce this exact message. A placeholder URN: the agent invented an id instead of listing first, and LinkedIn answers with this rather than a clear bad-argument error. A deleted post: the target is gone, which the message says explicitly. And a permission boundary: you cannot comment on a post the connected account cannot see, which includes company-page content your account has no access to. The fix is to start from a real id: linkedin_get_my_info for your profile, or a URN taken from a post URL you provide, then act on the URN that came back. The prefix is not interchangeable: production errors show both urn:li:share and urn:li:ugcPost forms rejected for the same content, so the URN should come from a prior tool response or the post's own URL rather than being built by hand.

Ask the agent to diagnose it:

Show me where the post URN came from. If it was not returned by a prior LinkedIn tool call or taken from the post URL I gave you, stop and run a list or search tool first.

LinkedIn caps post text at 3,000 characters at the API level, and the validation error is literal about which argument broke. Agents writing long-form posts will hit this on their own, so the fix is to ask for the draft to be trimmed to 3,000 characters or to be published as an article instead, which has its own tool, linkedin_create_article_or_url_share. The error is not a connection problem and reconnecting will not help.

Ask the agent to diagnose it:

Retry the post with commentary trimmed to 3,000 characters or less, keeping the first paragraph intact. Tell me the final length.

Try reconnecting LinkedIn 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 LinkedIn: link your LinkedIn account once in the browser and Hermes Agent can call the LinkedIn API through 10 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 LinkedIn 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 LinkedIn 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 LinkedIn working from chat.

Not for this connection. Building the integration yourself means registering an app in the LinkedIn Developer Portal, requesting the Share on LinkedIn product, waiting for approval, and handling token exchange. ClawLink connects through an app that is already registered, so what you do is sign in with your LinkedIn account and approve the grant. Nothing about that changes LinkedIn's own permission model: the connection still gets the scopes listed on the consent screen, and you can still revoke them later from LinkedIn's settings.

Not through this connection. The tool linkedin_create_linked_in_post accepts an organization author URN, but posting as an organization needs the w_organization_social scope, which this hosted connection does not request; the connection's scope list is openid, profile, w_member_social, and email. Company-page posting also sits behind LinkedIn's Community Management API and a page-admin role on your account. If org posting is the whole goal, this page is the wrong surface, and reconnecting from the dashboard will not fix it, because the scope list is fixed on the app side.

The risk warnings about AI on LinkedIn mostly describe a specific failure mode: a tool that logs in for you with your password or simulates a browser to act like a person. That is the automation pattern LinkedIn bans. This connection is the other model. Nothing stores your password, the agent acts under a grant you approved on LinkedIn's consent screen, the grant covers the 10 tools on this page rather than an open API session, and you can revoke it from your LinkedIn settings without involving us. The write tools are individually named: linkedin_create_linked_in_post, linkedin_create_comment_on_post, and linkedin_delete_linked_in_post, and posting carries a confirmation step in the flow.

Because the token never touches you. LinkedIn issues an access token to the connection, and ClawLink's side stores and refreshes it, so there is nothing for you to paste, rotate, or debug. The claim is about what you manage, not about what exists. An AI Overview described this setup as token-free, which overstates it: the token exists, it just lives server-side rather than in a config file next to your agent.

Run hermes clawlink test to confirm the pairing, then ask Hermes for a live read, for example "Use LinkedIn to show me my profile". linkedin_get_my_info returns the authenticated member's profile with no arguments, so a profile back in chat means the connection works. From there the natural next step is a real write under your control: draft a post and have the agent call linkedin_create_linked_in_post with visibility set to CONNECTIONS rather than PUBLIC.

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.