Your build starts here
Replace bracketed details with your own goal, context, and checks. Keep the first build small enough to inspect.
Prepare a change record for this prompt revision: [baseline and candidate]. The intended behavior change is [change], and the available evaluation evidence is [actual results or none].
Record the prompt identifiers or source revisions, input fixture version, dynamic template inputs, retrieval corpus or source references, relevant model configuration, and tool setup. Identify which of those changed and which stayed fixed. Keep private values out of the record.
Summarize representative improvements, regressions, and unrun checks using the actual evidence. If no evaluation ran, provide the next concrete comparison rather than inventing a score. State the selection decision and its limits.
Return a concise release note plus a machine-readable metadata outline if useful. Include how to restore the earlier prompt and associated configuration. A prompt label alone should not be presented as enough to reproduce a result whose other material inputs are unknown.Make this prompt yours
Use when prompts affect recurring application behavior. Keep records aligned with the project’s source and release workflow.
You might also hear
- prompt versioning
- prompt changelog
- reproducible prompt release
The docs behind it
These references explain the underlying concepts and APIs. The prompt templates and project plans are original examples to adapt to your own setup.
Research sources reviewed .
Editorial review · October 28, 2026
Supported: Original, unexecuted release-record template. OpenAI supports code-managed prompt revisions and regression checks; the requested provenance makes comparisons inspectable without claiming unchanged external inputs.