Using Claude Code instead?
The terminal client takes a
tgcc_ bearer key and is one command — and one
committed .mcp.json gives the whole team the same connection.Add the connector
Custom connectors are added under Settings → Connectors → Add custom connector. Paste the endpoint:1
Paste the URL
Claude fetches
/.well-known/oauth-protected-resource to learn which
authorization server issues tokens for this resource and which scopes it
understands. Nothing to configure — the server advertises it.2
Sign in and consent
You are sent to the dashboard, then shown a consent screen naming Claude and
the exact scopes it is asking for. That screen is the decision point: it is
where you choose what a model may do with your workspace.
3
Enable it in a chat
The connector appears in the tools menu. Turn it on per conversation.
List the agents in my ImpelLabs workspace and tell me which are published.
Some organisations also see a Request headers field in that dialog, which
takes a
tgcc_ key directly as Bearer tgcc_… (the space after Bearer
matters). It is beta and not enabled everywhere — if you do not see it, use
OAuth, which is the supported path for every account.Which one to use
Two things that will surprise you
Some tools answer with a job id, not a result
Some tools answer with a job id, not a result
Extractions, reports, knowledge ingestions and jobs are queued, always. The
tool returns the job id and Claude has to poll before it can honestly say
the work finished. This is not a failure and it is not a timeout.
Uploading a document succeeds before it is readable
Uploading a document succeeds before it is readable
Upload returns success while parsing continues in the background. Asking for
the document’s text straight afterwards returns 409
document_not_processed until parsing finishes. Claude will retry if you
tell it to; it cannot know to wait unless the tool description says so, and
it does.Tools
What is callable, and the scope each one demands.
Claude Code
The CLI: a bearer key, project scope, and a shared
.mcp.json.
