microsoft / microsoft/win-dev-skills
Tracking: Address feedback from joefinapps.com blog on building Fluent Icon Finder
Nobody has claimed this yet.
- Dominant language
- C#
- Stars
- 444
- Forks
- 33
- Avg merge
- 2h 59m
- Merged PRs (30d)
- 11
Description
Background
Joe Finney used the winui skill (in autopilot mode) to build TheJoeFin/Win-Icon-Finder — a Fluent icon similarity search app — and wrote up the experience in Building Fluent Icon Finder with the WinUI Copilot skill.
His overall verdict: "I highly recommend it!" The core feature work came out well — the agent independently arrived at a non-trivial matching pipeline (Win2D pre-render at 64×64 → L2-normalized grayscale vectors → cosine similarity → binary cache for fast subsequent launches), and MVVM structure landed cleanly.
But the post identifies five concrete, actionable gaps in the operational layer — scaffolding, packaging, CI, signing, and code style — that we should close. This is the umbrella tracking issue for those gaps.
Sub-issues
- #98 —
winui-packaging: add production signing & distribution guidance (Azure Trusted Signing,.appinstaller, EV/OV certs) — highest impact - #99 —
winui-packaging: ship a tested GitHub Actions workflow file instead of an inline YAML snippet - #100 —
winui-dev-workflow: generate a.slnxsolution file during greenfield scaffold - #101 —
winui-dev-workflow+ analyzer: set<RuntimeIdentifiers>for multi-arch packaging by default - #102 —
winui-dev-workflow+winui-code-review: seed canonical.editorconfigand run style polish pass on greenfield apps
Cross-cutting themes
These aren't standalone work items, but they're the pattern behind why these gaps exist:
- The shipping story is the gap. Agents nail the code; humans still own the ship. The highest-leverage single investment we can make is closing
dotnet new→ signed, installable, updatable artifact end-to-end. (Drives #98.) - Every snippet we tell the agent to emit should be a real, tested file in this repo. The CI YAML failure is a foot-gun caused by ungrounded inline prose. If it lives as a file exercised by CI, it can't rot. (Drives #99; pattern applies more broadly.)
- Greenfield needs an explicit polish phase. Most gaps below cluster around "the moment after first build." A single greenfield-only step that drops in
.slnx,.editorconfig,<RuntimeIdentifiers>, and runs a style autofix would solve #100, #101, and #102 together. Worth considering whether to implement them separately or as one unified post-scaffold step.
Original post
https://joefinapps.com/2026/05/25/building-fluent-icon-finder-with-the-winui-copilot-skill/
Reference repository
https://github.com/TheJoeFin/Win-Icon-Finder
Acknowledgments
Thanks to Joe Finney for the detailed, candid write-up — exactly the kind of preview feedback that lets us prioritize correctly.
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start with sub-issues #98–#102 and inspect the referenced winui-packaging, winui-dev-workflow, and winui-code-review skills. Review the original blog post and reference repository for the operational gaps. Done means the five concrete gaps have been addressed, including tested workflow and packaging guidance, with the tracking checklist updated.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- azure, csharp, github-actions
- Domain
- build-system, developer-experience, devops, release, tooling
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Quiet
- Clarity
- Mostly clear
- Newbie friendliness
- 35/100