Skip to content

chore: release v0.7.96 - #3075

Merged
vanceingalls merged 2 commits into
mainfrom
release/v0.7.96
Aug 7, 2026
Merged

chore: release v0.7.96#3075
vanceingalls merged 2 commits into
mainfrom
release/v0.7.96

Conversation

@vanceingalls

Copy link
Copy Markdown
Collaborator

What

Stable release v0.7.96, covering everything since v0.7.94 — the last version actually on npm.

Headline is canary attribution: rollouts now record why each install landed in or out of a cohort, on both the CLI and Studio (#3074). Before this a forced HF_CANARY_* override and an ordinary cohort roll produced an identical assignment, so cohort changes could not be explained. Also ships identity-persistence classification on every telemetry event (#3065), a Studio fix for rooted timeline media in preview (#3061), and cancelable thumbnail generation (#2720).

Why this is 0.7.96 and not 0.7.95

v0.7.95 was never published. Its release PR (#3068) merged correctly at 17:12 UTC and the publish workflow fired — but that run has been stuck in_progress for hours and produced no tag and no npm artifact. npm latest is still 0.7.94, while main's package.json reads 0.7.95.

The cause is the ongoing GitHub incident: Actions is in major_outage (started 15:22 UTC, still unresolved). Webhook triggers are throttled, so pushes and PRs are not scheduling runs, and runners are being assigned jobs that no longer exist.

Because v0.7.95 has no tag, the generated notes correctly span v0.7.94...v0.7.96, so nothing from 0.7.95 is lost — its changelog entry is also already on main from #3068.

⚠️ Merge ordering matters

Do not merge this until v0.7.95 is settled. The publish workflow sets the npm latest dist-tag. If 0.7.96 publishes and the stuck 0.7.95 run later resumes, latest would move backwards to 0.7.95.

Suggested order once Actions recovers:

  1. Let the 0.7.95 run finish, or re-run its merged-PR event (the documented recovery path), or cancel it if it is unrecoverable.
  2. Confirm CI is green on mainfeat(core,cli): emit the canary decision reason alongside the assignment #3074 merged with no matrix coverage at all (only WIP plus a skipped Mintlify), because of the same outage.
  3. Then merge this.

Verification

Local suites on the merged content: core 1674, cli 2475, studio 3480, studio-server 416. Lint clean. The canary-reason change was mutation-tested at all three layers — core helper, CLI binding, Studio binding.

This PR itself will show no checks until Actions recovers.

🤖 Generated with Claude Code

@mintlify

mintlify Bot commented Aug 6, 2026

Copy link
Copy Markdown

Preview deployment for your docs. Learn more about Mintlify Previews.

Project Status Preview Updated (UTC)
hyperframes 🟢 Ready View Preview Aug 6, 2026, 9:55 PM

💡 Tip: Enable Workflows to automatically generate PRs for you.

The pull_request webhook for this branch was dropped during the GitHub
Actions incident (major_outage 15:22-00:0x UTC), so #3075 received only the
WIP check and a skipped Mintlify. Dropped webhooks are not replayed, and
ci.yml has no workflow_dispatch, so an empty commit is the only way to fire
the event without closing the PR.

No content change: the release commit 026e694 is unmodified beneath this.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
@vanceingalls

Copy link
Copy Markdown
Collaborator Author

Ready for review — all checks terminal: 55 success, 0 failing, 5 skipped (path-filtered). mergeStateStatus is BLOCKED only on the pull_request ruleset's approval requirement.

Context worth having before approving, since two of the last three releases went wrong:

This ships 0.7.94 → 0.7.96. v0.7.95 is permanently skipped. Its release PR (#3068) merged correctly at 17:12 UTC and fired the publish workflow, but that run was caught by the GitHub Actions incident (major_outage, 15:22–00:0x UTC). It is now a phantom: zero jobs, cancel and force-cancel both return HTTP 500, re-run refuses with "already running". It cannot execute anything, so it cannot publish — which also means it cannot move the npm latest tag backwards. publish.yml has no concurrency group, so it will not block this release either.

Nothing is lost: since v0.7.95 was never tagged, the generated notes correctly span v0.7.94...v0.7.96, and 0.7.95's changelog entry is already on main from #3068.

The empty commit ae72e8c09 is deliberate. The original pull_request webhook for this branch was dropped during the outage, and dropped webhooks are not replayed. ci.yml has no workflow_dispatch, so an empty commit was the only way to fire the event without closing the PR. The release commit 026e6941a is unmodified beneath it, and no tag has been pushed — the workflow creates v0.7.96 from the merge SHA itself.

This is also the first CI the canary work has seen. #3074 merged during the outage with only WIP and a skipped Mintlify — no matrix at all. Those 55 green checks are the first real verification of the canary-reason emission on all three layers (core helper, CLI binding, Studio binding).

Merging publishes to npm under latest.

@vanceingalls
vanceingalls merged commit 066af73 into main Aug 7, 2026
60 checks passed
@vanceingalls
vanceingalls deleted the release/v0.7.96 branch August 7, 2026 00:37
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant