verification-before-completion
Blocks the agent from claiming work is complete, fixed, or passing until it has run verification commands and confirmed output. Evidence before assertions, always.
Anatomy
| Parent repo | obra/superpowers |
|---|---|
| Path | skills/verification-before-completion |
| Trigger conditions | "Use when about to claim work is complete, fixed, or passing, before committing or creating PRs" |
| Instruction format | yaml-frontmatter-md |
| Required tools | shell |
| Reusability | universal |
| Complexity | low |
| Ships scripts / examples / tests / boundaries | no / yes / no / yes |
Failure modes
Agent can rationalize that a task 'does not count' as a completion claim; the skill's red-flag table exists precisely to counter this, but enforcement is behavioral, not mechanical.
Verdict
The single highest-leverage skill in the index: one page of instructions that attacks the most expensive agent failure mode (declared-done-but-broken). Universal, tool-light, and honest about its own bypass risk.
Related skills
Forces root-cause investigation before any fix: reproduce, hypothesize one cause at a time, test the hypothesis, then patch. Bans…
Requires a failing test before implementation code for any feature or bugfix. The loop is red, green, refactor, with the agent forbidden…
Frequently asked questions
What does the verification-before-completion skill do?
Blocks the agent from claiming work is complete, fixed, or passing until it has run verification commands and confirmed output. Evidence before assertions, always.
When does verification-before-completion trigger?
Its stated trigger: "Use when about to claim work is complete, fixed, or passing, before committing or creating PRs". Trigger precision is a scored reliability dimension.