release-process
Skill2 starsChanged 33 days agoMITView on GitHub
Cuts a release with basicly release, pushes it and confirms it published. Use when cutting, tagging or checking a release.
- Commits and pushes
14 other projects use a version of this, 1 of them edited.
Adopted by
Most recently taken up by
- github/squadtheir own version, first seen Nov 8, 2026
- niksavis/basiclyedited, first seen Nov 7, 2026
- google/capsemtheir own version, first seen Oct 30, 2026
- microsoft/github-vteam-industry-demostheir own version, first seen Oct 25, 2026
- theultimate-dev/skillstheir own version, first seen Oct 11, 2026
- padamson/playwright-rusttheir own version, first seen Oct 8, 2026
Other versions14 projects use a version of this.
Word for word · 0
None found.
Edited · 1
- niksavis/basicly★ 271% same
Wrote their own version · 13
- fossasia/visdom★ 10ksame name
- github/gh-aw★ 5.6ksame name
- github/squad★ 3.4ksame name
- pluk-inc/markdown-preview★ 2.8ksame name
- pluk-inc/md-preview.app★ 2.8ksame name
- GoogleCloudPlatform/cloud-foundation-fabric★ 2.1ksame name
- lichtblick-suite/lichtblick★ 1.2ksame name
- google/capsem★ 179same name
- padamson/playwright-rust★ 166same name
- Azure/apiops-cli★ 32same name
3 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository54 other files it gives its agents.
- AGENTS.mdAGENTS.md
- .claude/CLAUDE.mdCLAUDE.md
- .github/copilot-instructions.mdCopilot instructions
- best-practices-auditSkill
- catalog-authoringSkill
- cli-toolsSkill
- conventional-commitsSkill
- decompose-planSkill
- falsify-firstSkill
- find-skillsSkill
- harness-clientSkill
- harness-loopSkill
- interface-factsSkill
- no-commentsSkill
- nodeSkill
- plain-englishSkill
- pythonSkill
- release-processSkill
- repair-in-placeSkill
- retention-probeSkill
- root-causeSkill
- session-finishSkill
- skill-creatorSkill
- test-disciplineSkill
- tier-injectionSkill
- validate-as-consumerSkill
- work-trackerSkill
- worktree-isolationSkill
- wslSkill
- best-practices-auditSkill
- catalog-authoringSkill
- cli-toolsSkill
- conventional-commitsSkill
- decompose-planSkill
- falsify-firstSkill
- find-skillsSkill
- harness-clientSkill
- harness-loopSkill
- interface-factsSkill
- no-commentsSkill
- nodeSkill
- plain-englishSkill
- pythonSkill
- repair-in-placeSkill
- retention-probeSkill
- root-causeSkill
- session-finishSkill
- skill-creatorSkill
- test-disciplineSkill
- tier-injectionSkill
- validate-as-consumerSkill
- work-trackerSkill
- worktree-isolationSkill
- wslSkill
The file
We split this skill into the rules it gives an agent
5 rules. 0 are common, 4 are distinctive, and 1 is very specific to this project.
0
Common
Most of the 14 edited versions we read keep these.
4
Distinctive
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 26
If either is missing, stop and follow the repository's own release document.
- Very specific: line 32
Write a summary under ## [Unreleased], above its first ### heading.
- Distinctive: line 64
Keep publication inside the version and scope the user requested.
- Distinctive: line 68
Run the two commands that the release prints:
- Distinctive: line 77
When a tag push is blocked by the Claude auto-mode classifier, read auto-mode.
“Common” is measured: kept by at least 6 in 10 of the 14 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
<!-- Generated by `basicly skills-build` from skill.yaml. Do not edit; edit the source. -->Release Process
Cut a release with basicly release, which does every step up to the annotated tag. You choose the version, write the summary and push.
Rules
- Confirm that this repository releases with
basicly releasefirst. The skill ships to every repository that installs basicly, also to one with its own release process. - Never do by hand a step that
basicly releasedoes. The command checks every precondition before the first write and restores the tree when a later step fails. - Write the notes before the cut. The release commit and the tag are one step, so a later commit never reaches the tagged notes.
- Never clear a refusal by a manual run of the named tool. That records a witness for a check that the gate did not see pass.
- Never delete a changelog fragment to clear a refusal. A release note must not disappear without notice.
- The command never pushes. Push only when the user release request includes publication.
1. Check that the repository uses it
ls changelog.d/ 2>/dev/null; grep -n "## \[Unreleased\]" CHANGELOG.mdIf either is missing, stop and follow the repository's own release document.
2. Write the notes
- Each change leaves its entry in
changelog.d/<record-id>.<category>.md. The categories areadded,changed,deprecated,removed,fixedandsecurity. One file per change means that two lanes never collide on one anchor. - A hand-written body under
## [Unreleased]also publishes, beside the fragments. - Write a summary under
## [Unreleased], above its first###heading. Write one paragraph on what the release changes for a user, then themed bullets. Name the record id of a fragment in each highlight.
The release folds each fragment into ## [Unreleased] under its Keep a Changelog heading, sorted by category and then by file name. It moves that body into the dated ## vX.Y.Z - <date> section, deletes the fragments in the same commit, and leaves [Unreleased] empty.
3. Wait for green CI
Cut only from a commit whose CI run on the release branch passed. A red or still running build is a stop.
4. Choose the version and run the dry run
Choose major, minor or patch yourself. The command refuses a version that does not move forward, but it does not choose one.
uv run basicly release <version> --issue <record-id> --dry-runThe dry run writes nothing and reports every refusal at once, each with the file or value it names. refusals lists them.
A page under docs/tutorial/ that names the current version as the one it was recorded against is rewritten to the new version by the release. A page that names any other version refuses: walk it against a build of this checkout and fix it first.
The dry run assembles the release in a temporary checkout and runs its actual commit hooks. It reports failing gates or the verify runner's gate count without changing this checkout. It copies the prepared Python environment and uses the same hook configuration. Find out whether the tag workflow runs the full checks before publication; when it does not, run required full checks on the release commit before pushing the tag.
5. Cut the release
uv run basicly release <version> --issue <record-id>It bumps __version__, regenerates the projections, rewrites the vX.Y.Z install pins in README.md and the other pin files, assembles the changelog, commits and creates the annotated tag. Add --date YYYY-MM-DD to set the changelog date. The default is today.
The user release request authorizes the requested cut and publication. No L-level grant or mandatory budget is needed.
Required release gates must still pass. Use --shipping <node> when the verified shipping node differs from the root.
Keep publication inside the version and scope the user requested.
6. Push and confirm
Run the two commands that the release prints:
git push origin main
git push origin vX.Y.ZWhen a workflow publishes from the tag push, it reads CHANGELOG.md from the tagged commit. Confirm that the workflow run is green and that the release page matches the summary. Never edit the release page after publication. A page that disagrees with its changelog section is a defect.
When a tag push is blocked by the Claude auto-mode classifier, read auto-mode.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

