Troubleshooting Workflows
Troubleshooting Workflows
Inspect partial state, identify the failed boundary, and choose the smallest safe recovery action.
custy --debugCategory
Quick Command
observe -> inspect state -> correct cause -> recover narrowly -> verifyFirst Principle
A failed Custy pipeline is not necessarily an untouched pipeline. Earlier steps may have generated or edited files, pruned backups, staged changes, created a commit or tag, or updated one or more remotes.
Recovery Method
Diagnose Before Changing State
Evidence Checklist
git status --short --branch
git diff
git diff --staged
git log -3 --decorate --stat
git tag --points-at HEAD
git branch --all
git remote -vBefore You Begin
- Read the first errorRequired
Later messages can be consequences; begin with the step that raised the failure.
- Inspect generated filesRequired
Check version files, editable artifacts, CHANGELOG.md, and message files.
- Inspect backupsRequired
Confirm whether new copies were created or older copies were pruned.
- Inspect Git objectsRequired
Determine whether staging, commit, tag, or branch deletion already occurred.
- Inspect every destinationRequired
For multi-remote work, verify each remote independently.
Choose Recovery by Last Completed Effect
Common Boundaries
Validation Only
Files Changed
Commit Exists
Tag Exists
Some Remotes Succeeded
Finalization Failed
Improve Diagnostic Output
Global options belong before the command group.custy --debug run release
custy --log-level DEBUG run dev
custy --debug cleanup brancheses --prefix feature/
Use debug output to understand resolution and execution, but do not include secrets, tokens, or sensitive remote credentials when sharing logs.
Escalate When State Is Unclear
Stop and ask a teammate or repository administrator before deleting commits or tags, force-pushing, changing shared branches, restoring remote branches, or rewriting published release history. Preserve logs and current state while the recovery decision is reviewed.