Buddy: debugging a failed execution - read failed action logs before retrying

[jtsternberg/buddy-cli]: the order is buddy executions failed for the failed action logs, then buddy executions show with the logs flag for all actions (the verbose flag surfaces errors from skipped actions), then failed with the analyze flag for error pattern analysis. For targeted debugging, use executions actions to find the specific action's hex ID, then action-logs to fetch just that one action, which is faster than pulling all logs sequentially. URL execution IDs are hex hashes and the CLI resolves them automatically. Only suggest fixes or retry after reading the logs.

Context: The fastest Buddy debugging loop reads the failed action's own logs first, then the full log, then analyzes the error pattern. Never guess before reading logs.

Source: The fastest Buddy debugging loop reads the failed action's own logs first, then the full log, then analyzes the error pattern. Never guess before reading logs.

Verified resolution: [jtsternberg/buddy-cli]: the order is buddy executions failed for the failed action logs, then buddy executions show with the logs flag for all actions (the verbose flag surfaces errors from skipped actions), then failed with the analyze flag for error pattern analysis. For targeted debugging, use executions actions to find the specific action's hex ID, then action-logs to fetch just that one action, which is faster than pulling all logs sequentially. URL execution IDs are hex hashes and the CLI resolves them automatically. Only suggest fixes or retry after reading the logs.

Source URL: https://github.com/jtsternberg/buddy-cli/blob/HEAD/claude-plugin/skills/troubleshooting-pipelines/SKILL.md