Skip to content

Latest commit

 

History

History
35 lines (23 loc) · 935 Bytes

File metadata and controls

35 lines (23 loc) · 935 Bytes

Explainability

Purpose

KOMU should explain its decisions instead of asking the operator to trust hidden reasoning.

Current explain features

  • komu explain last
  • komu explain plan <id>

These commands summarize:

  • why a playbook was chosen
  • why a provider was selected
  • why commands were considered risky
  • what the plan intends to change
  • what rollback guidance exists
  • what failed and what repair direction was prepared

Data source

Explain output is built from stored task history:

  • reviewed plan data
  • execution results
  • repair attempts

This keeps explanations auditable instead of depending on transient terminal output.

Current limits

  • explanations summarize recorded facts; they do not replay hidden model reasoning
  • older history entries created before this hardening pass may contain less review detail
  • rollback explanations are only as detailed as the originating playbook or plan data