| name | orchestrator-release |
| description | Prepares and validates GPT-RAG Orchestrator releases. Use for VERSION, changelog entries, release branches, tags, GitHub Release notes, and synchronized develop follow-up. |
GPT-RAG Orchestrator release
Read .github/copilot-instructions.md and
.github/instructions/release.instructions.md completely before changing a
release artifact.
- Determine the intended semantic version and create
release/X.Y.Z from
develop.
- Set root
VERSION to X.Y.Z without a v prefix.
- Replace the single
[Unreleased] heading with
## [vX.Y.Z] - YYYY-MM-DD; do not leave [Unreleased] on the release
branch.
- Verify the branch,
VERSION, changelog, intended tag, and GitHub Release
title are synchronized, accounting for their required prefix differences.
- Keep only release preparation on the branch and target the pull request to
main.
- Validate the exact commit with the relevant pytest, frontend, deployment,
contract, and Copilot asset checks.
- Keep changelog, release notes, and published operator documentation
consistent and free of secrets or private Azure environment names.
- Use exactly
vX.Y.Z for both the Git tag and GitHub Release title.
- After pushing the release branch, restore one empty
[Unreleased] section
at the top of develop, commit it, and push develop.
- Re-fetch published release notes and verify headings, lists, tables, and
sanitized content.
Do not publish a tag, release, package, image, or production deployment
without explicit human approval. Report failed validation or version drift as
a blocker rather than filling gaps by assumption.
