Skip to content

Latest commit

 

History

History
44 lines (27 loc) · 1.15 KB

File metadata and controls

44 lines (27 loc) · 1.15 KB

Verification Discipline

Verification is the difference between a workflow that feels smart and a workflow that ships reliably.

The baseline rule

Claims should be backed by evidence proportional to the risk of the change.

Documentation changes

For documentation-heavy work, verification usually means:

  • links resolve
  • file paths are real
  • the structure is coherent
  • examples match the documented model

Code and config changes

For executable work, verification typically expands to:

  • diagnostics
  • targeted tests
  • builds
  • manual sanity checks

Runtime-sensitive claims

When a workflow depends on production reality, local confidence is not enough.

That is when runtime signals, issue trackers, logs, traces, or tools like Sentry matter.

Questions worth asking before claiming done

  • What evidence do I have right now?
  • Is it local evidence or runtime evidence?
  • What could still be wrong even if the docs or tests look clean?
  • Am I reporting certainty or just momentum?

Why this playbook cares so much

AI accelerates drafting. It also accelerates plausible mistakes.

Verification is how you keep the speed while reducing self-deception.