Add agentic maintenance as local Copilot skills#1576
Open
kotlarmilos wants to merge 2 commits into
Open
Conversation
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: dd0486dd-9ff6-486b-aaa1-c4db25d01588
Contributor
There was a problem hiding this comment.
Pull request overview
This PR introduces repository-local Copilot “skills” under .github/skills/ to run TorchSharp maintenance tasks (CI failure scanning, scanner feedback, and LibTorch bump prep) locally with explicit approval gates for any GitHub writes.
Changes:
- Add a
ci-scanskill plus detailed reference methodology (playbook + shared instructions) for locally scanning recent AzDO CI failures and drafting up to three tracking issues. - Add a
ci-scan-feedbackskill to audit scanner output/maintainer feedback and propose prompt/method improvements. - Add a
libtorch-bumpskill to locally detect new PyTorch releases and prepare safe patch-level LibTorch version updates.
Show a summary per file
| File | Description |
|---|---|
| .github/skills/libtorch-bump/SKILL.md | Documents a local, approval-gated workflow for LibTorch patch bumps while preserving macOS x64 override pins. |
| .github/skills/ci-scan/SKILL.md | Defines the local CI-scan skill entrypoint and its approval/validation rules. |
| .github/skills/ci-scan/references/playbook.md | Adds the detailed CI-scan playbook steps for selecting builds, extracting signatures, deduping, and drafting issues. |
| .github/skills/ci-scan/references/ci-scan.instructions.md | Provides shared, reusable methodology sections (classification, dedup, sanitization, template, rubric). |
| .github/skills/ci-scan-feedback/SKILL.md | Defines the feedback/audit workflow for improving the ci-scan skill based on real issues and signals. |
Copilot's findings
Tip
Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
- Files reviewed: 5/5 changed files
- Comments generated: 2
matouskozak
approved these changes
Jul 20, 2026
The per-leg signature-source table duplicated the content-driven Failure classification in ci-scan.instructions.md, which derives build break, test failure, and infra noise from the failing task type and diagnostic line rather than from leg names. The infra-noise cases it pre-labeled, signing, publishing, and large libtorch downloads, are already listed there, so removing the table changes no output and drops a hard-coded leg list that would need maintenance as CI legs change. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: ccd096a7-a2c0-4bed-8e67-089c46184e52
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Description
This adds the TorchSharp CI scanner, scanner feedback, and libtorch bump automation as repository-local Copilot skills under
.github/skills/. Local execution uses the operator's authenticated tools and keeps GitHub writes behind explicit approval while preserving the scanner methodology as progressive-disclosure references. This avoids PAT-pool and scheduled workflow configuration until repository billing and authentication are available. The deferred CI implementation remains in #1575 for a later follow-up.