release-create-tracker-issue
Generate (and optionally open) a PyTorch release tracker / cherry-pick tracking issue from a release announcement, like…
npx skills add https://github.com/pytorch/test-infra --skill release-create-tracker-issueCreate Release Tracker Issue
Generates the body for a PyTorch release tracker issue (the cherry-pick tracking issue cut alongside a release branch, like https://github.com/pytorch/pytorch/issues/180506) from a release announcement that contains the milestone (key dates) schedule, then opens the issue in pytorch/pytorch.
When to use this skill
Use when the user asks to:
- Create a release tracker / release tracking issue for a PyTorch release
- Create the cherry-pick tracking issue for a newly cut release branch
- Turn a "PyTorch release X.Y key dates" announcement into a tracker issue
Instructions
Step 1: Determine parameters
Collect these parameters. Most can be derived; ask only for what is missing.
| Parameter | Description | Example | How to obtain |
|---|---|---|---|
| Release announcement | URL or pasted text of the "key dates" post | https://dev-discuss.pytorch.org/t/pytorch-release-2-13-key-dates/3390 | From the user |
| Release version | Full version being released | 2.13.0 | From the user or announcement title |
| Branch version | MAJOR.MINOR of the release branch | 2.13 | Derived from release version (drop the patch) |
| Previous minor | The most recent prior minor release | 2.12 | Branch minor minus 1 (e.g. 2.13 → 2.12) |
| Release managers | GitHub handles with cherry-pick dispensation | @atalman, @malfet | Default @atalman, @malfet unless told otherwise |
Step 2: Extract milestone dates from the announcement
If given a URL, fetch it (the dev-discuss forum is not a GitHub URL, so use WebFetch, not gh):
WebFetch(url=<announcement_url>, prompt="Extract the M3 (release branch cut), M4 (release branch finalized / feature classifications), M4.1 (tutorial drafts submission deadline), M5 (external-facing content finalized), and M6 (release day) dates verbatim, preserving any 'week of' prefix.")
If given pasted text, parse the dates directly from it.
Map the announcement milestones to the tracker fields. Keep the date format exactly as written in the announcement (e.g. DD/MM/YY, and keep a week of prefix if present):
| Tracker field | Source milestone |
|---|---|
M3_DATE | M3 — Release branch cut |
M4_DATE | M4 — Release branch finalized / feature classifications published |
M4_1_DATE | M4.1 — Tutorial drafts submission deadline |
M5_DATE | M5 — External-Facing Content Finalized |
M6_DATE | M6 — Release Day |
PHASE_CUTOFF | The M4 date. If M4 is written as week of 22/6/26, use the bare date 22/6/26 for the phase cutoff. |
If a milestone is missing from the announcement, ask the user rather than guessing.
Step 3: Generate the issue body
Fill the template below. Replace every {PLACEHOLDER}:
{VERSION}→ full release version (e.g.2.13.0){BRANCH}→ branch version (e.g.2.13){PREV_MINOR}→ previous minor (e.g.2.12){PHASE_CUTOFF},{M3_DATE},{M4_DATE},{M4_1_DATE},{M5_DATE},{M6_DATE}→ dates from Step 2{RELEASE_MANAGERS}→ e.g.@atalman, @malfet
We cut a [release branch](https://github.com/pytorch/pytorch/tree/release/{BRANCH}) for the {VERSION} release.
Our plan from this point is roughly:
* Phase 1 (until {PHASE_CUTOFF}): work on finalizing the release branch
* Phase 2 (after {PHASE_CUTOFF}): perform extended integration/stability/performance testing based on Release Candidate builds.
This issue is for tracking cherry-picks to the release branch.
## Release dates
* M3: Release branch cut ({M3_DATE})
* M4: Release branch finalized, Announce final launch date, Feature classifications published ({M4_DATE}) - Final RC is produced.
* M4.1: Tutorial drafts submission deadline ({M4_1_DATE})
* M5: External-Facing Content Finalized ({M5_DATE})
* M6: Release Day ({M6_DATE})
## Cherry-Pick Criteria
**Phase 1 (until {PHASE_CUTOFF}):**
Only low-risk changes may be cherry-picked from main:
1. Fixes to regressions against the most recent minor release (e.g. {PREV_MINOR}.x for this release; see [module: regression issue list](https://github.com/pytorch/pytorch/issues?q=is%3Aissue+is%3Aopen+label%3A%22module%3A+regression%22+))
2. Critical fixes for: [silent correctness](https://github.com/pytorch/pytorch/issues?q=is%3Aissue+is%3Aopen+label%3A%22topic%3A+correctness+%28silent%29%22), [backwards compatibility](https://github.com/pytorch/pytorch/issues?q=is%3Aissue+is%3Aopen+label%3A%22topic%3A+bc-breaking%22+), [crashes](https://github.com/pytorch/pytorch/issues?q=is%3Aissue+is%3Aopen+label%3A%22topic%3A+crash%22+), [deadlocks](https://github.com/pytorch/pytorch/issues?q=is%3Aissue+is%3Aopen+label%3A%22topic%3A+deadlock%22+), (large) [memory leaks](https://github.com/pytorch/pytorch/issues?q=is%3Aissue+is%3Aopen+label%3A%22topic%3A+memory+usage%22+)
3. Critical fixes to new features introduced in the most recent minor release (e.g. {PREV_MINOR}.x for this release)
4. Test/CI fixes
5. Documentation improvements
6. Compilation fixes or ifdefs required for different versions of the compilers or third-party libraries
7. Release branch specific changes (e.g. change version identifiers)
Any other change requires special dispensation from the release managers (currently {RELEASE_MANAGERS} ). If this applies to your change please write "Special Dispensation" in the "Criteria Category:" template below and explain.
**Phase 2 (after {PHASE_CUTOFF}):**
Note that changes here require us to rebuild a Release Candidate and restart extended testing (likely delaying the release). Therefore, the only accepted changes are **Release-blocking** critical fixes for: [silent correctness](https://github.com/pytorch/pytorch/issues?q=is%3Aissue+is%3Aopen+label%3A%22topic%3A+correctness+%28silent%29%22), [backwards compatibility](https://github.com/pytorch/pytorch/issues?q=is%3Aissue+is%3Aopen+label%3A%22topic%3A+bc-breaking%22+), [crashes](https://github.com/pytorch/pytorch/issues?q=is%3Aissue+is%3Aopen+label%3A%22topic%3A+crash%22+), [deadlocks](https://github.com/pytorch/pytorch/issues?q=is%3Aissue+is%3Aopen+label%3A%22topic%3A+deadlock%22+), (large) [memory leaks](https://github.com/pytorch/pytorch/issues?q=is%3Aissue+is%3Aopen+label%3A%22topic%3A+memory+usage%22+)
Changes will likely require a discussion with the larger release team over VC or Slack.
## Cherry-Pick Process
1. Ensure your PR has landed in master. This does not apply for release-branch specific changes (see Phase 1 criteria).
2. Create (but do not land) a PR against the [release branch](https://github.com/pytorch/pytorch/tree/release/{BRANCH}).
<details>
```bash
# Find the hash of the commit you want to cherry pick
# (for example, abcdef12345)
git log
git fetch origin release/{BRANCH}
git checkout release/{BRANCH}
git cherry-pick -x abcdef12345
# Submit a PR based against 'release/{BRANCH}' either:
# via the GitHub UI
git push my-fork
# via the GitHub CLI
gh pr create --base release/{BRANCH}
```
You can also use the `@pytorchbot cherry-pick` command to cherry-pick your PR. To do this, just add a comment in your merged PR. For example:
```
@pytorchbot cherry-pick --onto release/{BRANCH} -c docs
```
(`-c docs` - is the category of your changes - adjust accordingly):
For more information, see [pytorchbot cherry-pick docs](https://github.com/pytorch/pytorch/wiki/Bot-commands#cherry-pick).
</details>
3. Make a request below with the following format:
```
Link to landed trunk PR (if applicable):
*
Link to release branch PR:
*
Criteria Category:
*
```
1. Someone from the release team will reply with approved / denied or ask for more information.
2. If approved, someone from the release team will merge your PR once the tests pass. **Do not land the release branch PR yourself.**
**NOTE: Our normal tools (ghstack / ghimport, etc.) do not work on the release branch.**
Please note HUD Link with branch CI status and link to the HUD to be provided here.
[HUD](https://hud.pytorch.org/hud/pytorch/pytorch/release%2F{BRANCH})
cc @seemethere @malfet @pytorch/pytorch-dev-infra
### Versions
{VERSION}
Step 4: Review, then create the issue
- Show the rendered body to the user for confirmation. Creating a GitHub issue is an outward-facing action — do not create it until the user approves the content (or has clearly asked you to open it directly).
- Create the issue in
pytorch/pytorchwith the exact title and labels:- Title:
[v.{VERSION}] Release Tracker(note thev.prefix — e.g.[v.2.13.0] Release Tracker) - Labels:
release tracker,triaged
- Title:
gh issue create \
--repo pytorch/pytorch \
--title "[v.{VERSION}] Release Tracker" \
--label "release tracker" \
--label "triaged" \
--body-file release_tracker_body.md
If gh is unavailable, POST to https://api.github.com/repos/pytorch/pytorch/issues with {"title": ..., "labels": ["release tracker", "triaged"], "body": ...} using an authenticated token.
- Report the new issue URL back to the user.
Example usage
Example — Create the 2.13.0 tracker from the key-dates announcement:
Create a release tracker issue for 2.13.0 from https://dev-discuss.pytorch.org/t/pytorch-release-2-13-key-dates/3390
For that announcement the milestones resolve to:
| Field | Value |
|---|---|
| M3 (branch cut) | week of 8/6/26 |
| M4 (finalized) | week of 22/6/26 |
| M4.1 (tutorial drafts) | 30/6/26 |
| M5 (content finalized) | 1/7/26 |
| M6 (release day) | 8/7/26 |
| Phase cutoff | 22/6/26 |
| Previous minor | 2.12 |
Producing title [v.2.13.0] Release Tracker with labels release tracker, triaged.
Notes
- Phase cutoff = the M4 date. Both the "Phase 1 (until …)" / "Phase 2 (after …)" lines and the "Phase 1/2" cherry-pick criteria headers use the bare M4 date (strip a
week ofprefix for the cutoff, even though the M4 line in the Release dates section keeps it). - Preserve the announcement's date format verbatim (
DD/MM/YY). Do not reformat or convert dates. - The
release trackerlabel must already exist in the repo (it does inpytorch/pytorch). If creating in a repo without it, create the label first or drop it. - Keep the trailing
cc @seemethere @malfet @pytorch/pytorch-dev-infraline unless the user specifies different reviewers. - The release managers handles default to
@atalman, @malfet; update only if the user names different managers.