Loop 029
The Revolve versioned-experiment loop
A Revolve workflow that improves prompts, code, or configurations through checkpointed experiments whose scores remain comparable across sessions.
Ready-to-use prompt
Copy the loop
Use Revolve to improve a support prompt, code path, or testable subject. In revolve/, define the goal and [budget], freeze the tests and scoring, checkpoint the current version, and record a baseline. Each round, test one hypothesis; keep only a clear, regression-free win. If the evaluation changes, open a new revision and rerun the baseline. Ask before changing live files. Stop on success, no progress, a blocker, or exhausted budget. Return the best checkpoint, comparisons, rollback, and next action.
Verify / stop
The best Revolve checkpoint wins within one evaluation revision.
The incumbent and candidates have comparable recorded runs, accepted changes pass every guard, rollback is available, and live promotion has approval.
Context and guidance When to use it, steps, safety notes, and related loops
Use this when
Use Revolve to improve a prompt, policy, workflow, model configuration, code path, or dataset when experiments must remain comparable and resumable across sessions.
How to run it
- Create or resume revolve/, define the objective and permissions, freeze an evaluation revision, checkpoint the incumbent, and record its baseline.
- Choose one evidence-backed hypothesis, create a candidate checkpoint, and test it under the unchanged revision.
- Promote internally only on a meaningful guard-safe win; if the evaluation changes, open a new revision and rerun the incumbent.
- Stop on a named condition, and require explicit approval plus verification before changing live files.
Why it works
Revolve's revision boundaries prevent scores from different tests or rubrics from being compared as equivalent. Checkpoints and an internal-before-live promotion boundary keep long-running research resumable and reversible.
Implementation note
The source examples include improving CLI error messages, reducing image-export latency, tuning a support-assistant prompt, and hardening a parser. Replace the subject and metric, but keep the revision, checkpoint, and rollback discipline.