Execution errors
When a run fails, the timeline shows the step that stopped and the reason. This page translates the reasons.
The Retries? column says whether the platform tries again automatically, when the step has retries configured.
Identity and permission
| Reason | Meaning | Retries? | What to do |
|---|---|---|---|
| Service account not configured | The organization has no service account | No | See Service account |
| Service account disabled | It exists and is inactive | No | Reactivate it |
| Divergent identity | The run was created under another identity | No | Trigger it again |
| Authorization expired | The authorization lapsed mid-work | Yes | Usually resolves on its own |
| Authorization revoked or tampered with | The authorization is no longer valid | No | Trigger it again; if it persists, contact support |
| Capability outside the authorization | The step asked for something the version did not declare | No | Declare the resource and publish again |
| Skill not permitted | The organization's policy does not grant it | No | See Allowed resources |
AI steps
| Reason | Meaning | Retries? | What to do |
|---|---|---|---|
| Engine unavailable | Imaginne's engine did not answer | Yes | Usually resolves on its own |
| Policy refused | The engine refused the policy context | No | Review skills and resources |
| Required tool unavailable | A required tool does not exist or did not complete | No | See Allowed resources |
| Tool failed | A tool ended in failure, with no declared tolerance | No | Investigate the tool, or declare tolerance |
| Output does not match the declared format | A required field did not come back | No | Reduce the fields or make the objective more explicit |
| The AI asked a person for information | In an automatic run there is nobody to answer | No | Complete the input data |
| Work limit reached | Iterations or answer size at the ceiling | No | Raise the limit or reduce the scope |
| Partial result | What came back is not the whole answer | No | Reduce the task's scope |
| Stopped with no result | No conclusion was reached | No | Review the objective |
| Model error | A model failure | Yes | Usually resolves on its own |
| Context exhausted | The content exceeded what fits in the context | Yes | Reduce the data handed over |
| Unknown stop reason | The engine returned something the platform does not recognize | No | Contact support |
That last row is a deliberate rule: a newer engine must not make an agent declare success out of ignorance.
External calls
| Reason | Meaning | Retries? | What to do |
|---|---|---|---|
| Invalid action input | The parameters do not match what the action expects | No | Fix the step's configuration |
| Action blocked | Policy or network protection | No | See Calling an API |
| Step timeout | The step exceeded its deadline | Yes | Raise the deadline or investigate the other side |
Flow structure
| Reason | Meaning | Retries? | What to do |
|---|---|---|---|
| Unreadable definition | The run's frozen content is invalid | No | Publish again |
| Stuck flow | Nothing is ready and no Result was reached | No | Review the connections |
| Failure deciding a path | An error evaluating a connection's condition | No | Review the condition |
| Failure assembling a value | An expression could not be evaluated | No | Review the values used |
| Key conflict while combining | Two sources would produce the same key | No | Name the sources |
Approvals and waiting
| Reason | Meaning | What to do |
|---|---|---|
| Approval rejected | Someone rejected it | Nothing — it is a legitimate outcome |
| Approval expired | The deadline ran out with no decision | Raise the deadline or notify whoever decides |
Agent chains
| Reason | Meaning | What to do |
|---|---|---|
| Depth exceeded | The chain is longer than the ceiling | Shorten the chain |
| Recursion detected | An agent in the chain calls someone already in it | Redesign the flow |
| Called agent not found | No active version in that environment | Activate the agent in the same environment |
| Child run failed | The problem is in the called agent | Open the child run |
Time
| Reason | Meaning | What to do |
|---|---|---|
| Run expired | The total deadline ran out | Raise the timeout or reduce the work |
What the messages never contain
Every message goes through a cleanup before being stored: credentials, tokens, and database addresses are stripped. What remains is enough to diagnose, and nothing that should not circulate.
Next steps
Was this page helpful?
Report a problem on this pageDo not send passwords, keys, tokens, or customer data.