Agent Workflows
Retry a failed assignment

Retry a failed assignment

Re-run a publish that ended in failed without recreating it.

User intents

"My upload failed — try again." "Retry all my failed posts from today."

Required information

  • The assignmentId(s) of the failed assignment(s).

Preconditions

  • The assignment is in failed state.
  • The underlying cause is fixed (e.g. token re-authorized, credits topped up).

Tool sequence

Find failed assignments

list_assignments with status: failed (optionally filter by connectionId or date).

Inspect the failure

get_assignment — read the error detail to understand why it failed before retrying.

Fix the root cause

If the token expired, get_oauth_connect_url; if the payload was invalid, get_platform_capabilities and (for not-yet-published) update_assignment.

Retry

retry_assignment with the assignmentId. It re-queues the same assignment.

⚠️

Don't retry blindly in a loop. Read the failure reason first — retrying an expired-token or insufficient-credits failure without fixing the cause will just fail again.

REST alternative

See the retry endpoint in the Assignments reference.

Recovery paths

  • insufficient_credits — top up before retrying (retry consumes credits again). See the error catalog (opens in a new tab).
  • Token expired — re-authorize, then retry.
  • Still failing after a valid retry — surface the platform error to the user.

Example prompt → result

"List failed assignments from today and retry the ones that failed due to a temporary error."

Result: each eligible failed assignment is re-queued and moves back through the publish lifecycle.

Common mistakes

  • Creating a brand-new assignment instead of retrying (wastes the original record and credits).
  • Retrying without inspecting get_assignment first.

Related docs

Error Handling guide · Tool reference · Publish now