Get help
Most questions about a run answer themselves from the run’s own record — this page is the short route to that record, the questions everyone asks first, and where to go when self-serve isn’t enough.
The self-serve path
- Open the run. Its terminal reason and execution log carry most answers — Troubleshoot a run decodes the common ones.
- Check the queue. A run that seems stuck is often just waiting on a human.
- Check the audit log. For anything involving permissions, denials, or “who changed this”, the audit log is the org-wide answer.
- Ask an administrator. Policy, provider keys, and server enablement are Settings-side fixes; the error messages name which.
Frequent questions
Why is my run “Awaiting approval”? A tool it called is set to Approve. The owner resolves it from the run page — and has an hour before it expires and fails the run.
Why did my run fail exactly an hour after pausing? That was the expiry. The terminal reason reads Approval expired without resolution.
Why don’t I see any models to pick? Your organization hasn’t configured a provider key your account can use — an administrator adds one under Settings → Secrets.
Why can’t I connect a service I know exists? It isn’t enabled for your organization yet, or it only takes an organization-level credential. Both are visible to an administrator in the MCP Registry.
Why did the same agent answer differently for my teammate? Tools authenticate as the person the agent acts for — different credentials see different data.
Can I get notified when a run fails or needs approval? Not yet — there are no email or chat notifications today. Monitoring is the run list and the Approvals page.
Why was my run “Skipped”? Its previous run was still active when the slot fired — one run at a time is a guarantee, not a failure.
Report an issue or request a feature
Where your workspace has it configured, the Feedback button in the sidebar footer is the direct line — it reaches the team operating your Lunen instance. A useful report names the agent, links the specific run (the run page URL carries the run ID), and quotes the terminal reason verbatim. Feature requests travel the same route; “what I was trying to do” is more useful to the team than “what button I wish existed.”
Where the Feedback button isn’t present, your workspace administrators are the escalation path — they hold the audit access and vendor relationship a diagnosis needs.