Abbey AI Workflow-Friction Decision

July 25, 2026

Added an evidence-backed AI decision strategy for reducing costly recurring workflow friction.

Tags: Abbey Root • AI • Developer Toolkit • Planning

Abbey AI Workflow-Friction Decision

Summary

Added abbey ai decide workflow-friction, a metadata-driven strategy for finding repeated manual work, awkward handoffs, duplicate entry, and recurring verification burdens that warrant a bounded Abbey improvement.

The strategy requires evidence of recurrence, separates recommendation confidence from implementation confidence, and requires repository review before implementation.

Accomplishments

• Added the workflow-friction decision metadata, prompt, and result schema.
• Prioritized cumulative recurring friction over one-time annoyances.
• Distinguished evidence from assumptions and prohibited invented implementation details.
• Required a bounded, maintainable improvement with a retained human boundary.
• Classified the result as an Abbey command, standardized workflow, or local fix.
• Extended the shared terminal report for workflow-friction results.
• Added focused metadata-discovery and decision-contract regression coverage.
• Captured the new decision as completed Abbey AI backlog work.

Lessons Learned

Not every repeated inconvenience should become a command. A useful friction decision must also judge the proportionate shape of the improvement.

Planning documents can support a strong recommendation while revealing little about implementation. The decision makes that mismatch visible through separate confidence values and required repository review.

Next Steps

• Validate a completed local-model recommendation from the canonical Ubuntu environment.
• Evaluate the classification boundary through normal use before expanding the rubric.