/verification-loop

name: verification-loop
description: Use when a task spans more than a few steps or files — establishes a checkable definition of done, a verification cadence, and fresh-context verifier subagents before claiming completion.

Verification Loop

Fable 5 builds its own checking harness on long tasks. Reproduce that explicitly.

1. Define "done" before starting

Write down (in your todo list or a scratch note) the binary checks that prove the task is complete: which tests must pass, which command must produce which output, which behavior must be observable. If the user's request doesn't define them, derive them and state them in one sentence.

2. Verify on a cadence, not at the end

After each meaningful increment (a function implemented, a module refactored), run the relevant real check immediately — the actual test command, the actual build. Never batch all verification to the end of a long task; errors compound.

3. Fresh-context verification before completion claims

Before declaring a substantial task complete, dispatch a verifier subagent (`modelharness:verifier` agent, or a general-purpose subagent given the `agents/verifier.md` instructions). Give it: the original specification, the definition-of-done checks, and the current diff (`git diff` output or changed file list). A fresh context catches what you have rationalized away. Self-review is the fallback only when subagents are unavailable.

4. Act on the verdict

If the verifier reports gaps: fix them, re-run the checks, re-verify. Only report completion to the user when checks pass AND the verifier (or the checks themselves) confirm it. Report any unverified remainder explicitly.