Back to catalog / Agent operations

Classify agent failures by structured exception

Separate missing CLI, connection loss, process failure, and terminal model errors instead of retrying every exception.

Goal

Convert SDK errors into auditable failure classes and appropriate recovery actions.

Import CLINotFoundError, CLIConnectionError, ProcessError, ResultError, and CLIJSONDecodeError. [readme]
Catch concrete types; for ResultError, inspect subtype, terminal_reason, api_error_status, result, or errors. [readme]
Retry only transient connection failures with a cap; report terminal errors, configuration failures, and decode failures separately. [readme]

Simulate each exception class and assert distinct alert/recovery handling with no unbounded retries.

If an error escapes, inspect the exception hierarchy and preserve structured fields; do not log full prompts or credentials.

Python 3.10+, SDK; exception mapping can be unit-tested without a live model call.

Log the error class and necessary status fields; protect api_error_status under organizational policy.

Error classification does not establish business correctness; exception fields are tied to the pinned SDK version.

These values describe the upstream repository, not popularity or likes for an individual skill or playbook.

No metrics snapshot is available yet.

License: MIT