Update WinUI skills for WinApp CLI 0.6 (#145)

* Update skills for WinApp CLI 0.6

Adopt winapp new, find-ui, and project-mode run while removing the superseded bundled winui-search tool and simplifying BuildAndRun.ps1.

Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 418fd91b-b050-438c-b3d3-b26c00bdeeea

* Rely on WinGet for WinApp CLI

Remove the temporary standalone release fallback; WinApp CLI 0.6 will be published to WinGet before these skills ship.

Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 418fd91b-b050-438c-b3d3-b26c00bdeeea

* Fix skill regressions from the WinApp CLI 0.6 migration

Review of the WinApp CLI 0.6 update surfaced three regressions where
behavior previously enforced by tooling was dropped or demoted to prose.

winui-packaging: Step 1 had switched from the BuildAndRun.ps1 wrapper to
raw `dotnet build`, which does not load the bundled
Microsoft.WindowsAppSDK.Analyzers -- so release builds lost the analyzer
gate that development builds get. Restore the wrapper using the new flag
surface (`--no-launch` replaces the removed `-SkipRun`). Unlike the old
build-only path, `--no-launch` also registers a development package, so
note the `winapp unregister` step before installing the signed MSIX. The
GitHub Actions example stays on `dotnet build` -- the wrapper registers
packages and needs the plugin's local analyzer payload, neither of which
belongs on a runner.

winui-design: the skill absorbed sample and control discovery via
`winapp find-ui` when winui-search.exe was removed, but its description
never gained matching trigger phrases, so sample-search requests no
longer activate it.

winui-dev-workflow: the 0x80073CF9 troubleshooting row replaced an
automatic MAX_PATH guard that keyed the staging directory on a hash of
the output path and purged it before reuse. The replacement was keyed on
app name alone with no purge, so Debug and Release now collide in one
directory that a registered development package holds a live reference
to. Re-encode both invariants in the guidance.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: b05a748c-f1f9-4159-84d2-d66d401a6067

---------

Co-authored-by: Nikola Metulev <711864+nmetulev@users.noreply.github.com>
Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Jaylyn Barbee <51131738+Jaylyn-Barbee@users.noreply.github.com>
Copilot-Session: 418fd91b-b050-438c-b3d3-b26c00bdeeea
Copilot-Session: b05a748c-f1f9-4159-84d2-d66d401a6067

Author: 3 people

.github/ISSUE_TEMPLATE/bug-report.yml

@@ -32,7 +32,6 @@ body:
         - "skill: winui-session-report"
         - "skill: winui-wpf-migration"
         - "tool: winui-analyzer (Roslyn)"
-        - "tool: winui-search"
         - "tool: winmd-cli"
         - "area: plugin / install"
         - "area: docs"

.github/PULL_REQUEST_TEMPLATE.md

@@ -29,9 +29,9 @@ The `pr-target-policy` CI check enforces this.
 
 <!-- Check all that apply -->
 
-- [ ] Agent (`plugins/winui/agents/`)
+- [ ] Agent (`plugins/winui/agents/`, `plugins/winui/agent-plugin/com.github.copilot/agents/`)
 - [ ] Skill: <!-- name(s) -->
-- [ ] Tool: <!-- winui-analyzer / winui-search / winmd-cli -->
+- [ ] Tool: <!-- winui-analyzer / winmd-cli -->
 - [ ] Plugin metadata (`plugin.json`, `plugins/winui/`)
 - [ ] Repo-level docs / governance
 

.github/codeql/codeql-config.yml

@@ -6,4 +6,3 @@ paths-ignore:
   - '**/*Tests/*.cs'
   - '**/bin/**'
   - '**/obj/**'
-  - 'src/tools/winui-search/Data/**'

.github/dependabot.yml

@@ -45,20 +45,6 @@ updates:
       prefix: "deps"
       include: "scope"
 
-  # winui-search CLI (gallery + toolkit scenario search)
-  - package-ecosystem: "nuget"
-    directory: "/src/tools/winui-search"
-    schedule:
-      interval: "weekly"
-      day: "monday"
-    open-pull-requests-limit: 5
-    labels:
-      - "dependencies"
-      - "tool: winui-search"
-    commit-message:
-      prefix: "deps"
-      include: "scope"
-
   # winmd-cli (WinMD inspector)
   - package-ecosystem: "nuget"
     directory: "/src/tools/winmd-cli"

.github/skills/pr-review/SKILL.md

@@ -16,11 +16,12 @@ This repo is **not a regular C# product**. It ships:
   prompt + skill prompts (`SKILL.md` files). These are **Tier 3 instructions**
   that agents frequently ignore (see `dimensions/skill-tool-boundary.md`).
   Adding prose here is the *last resort*, not the first response to any problem.
-- Three **in-repo C# tools** under `src/tools/` — the WinUI 3 Roslyn analyzer,
-  `winmd-cli`, and `winui-search`. These are **Tier 1 enforcement** and the
-  preferred place to land behavior changes.
-- **Committed binary payloads** (analyzer DLL, `winui-search.exe`,
-  `Microsoft.WindowsAppSDK.Analyzers.targets`) inside `plugins/winui/agent-plugin/skills/`
+- Two **in-repo C# tools** under `src/tools/` — the WinUI 3 Roslyn analyzer
+  and `winmd-cli`. These are **Tier 1 enforcement** and the preferred place
+  to land behavior changes that belong in this repository.
+- **Committed analyzer payloads** (DLL and
+  `Microsoft.WindowsAppSDK.Analyzers.targets`) inside
+  `plugins/winui/agent-plugin/skills/`
   that must stay in sync with their sources. CI provenance jobs will fail
   the PR if they drift, but it's better to flag the drift in review.
 
@@ -114,14 +115,13 @@ focus. Common buckets in this repo:
 | `plugins/winui/agent-plugin/skills/<name>/references/` | skill-content (references discipline) |
 | `plugins/winui/agent-plugin/skills/<name>/*.ps1` (e.g. `BuildAndRun.ps1`, `Analyze-Session.ps1`) | tool-correctness, payloads-and-tests |
 | `plugins/winui/agent-plugin/skills/winui-dev-workflow/analyzer/` | payloads-and-tests (committed analyzer payload) |
-| `plugins/winui/agent-plugin/skills/winui-design/winui-search.exe` | payloads-and-tests (committed AOT exe) |
-| `plugins/winui/agents/winui-dev.agent.md` | skill-content, docs-and-manifests |
+| `plugins/winui/{agents,agent-plugin/com.github.copilot/agents}/winui-dev.agent.md` | skill-content, docs-and-manifests |
 | `plugins/winui/agent-plugin/plugin.json` | docs-and-manifests |
 | `.github/plugin/marketplace.json` | docs-and-manifests |
 | `src/tools/winui-analyzer/Microsoft.WindowsAppSDK.Analyzers/` | tool-correctness, payloads-and-tests |
 | `src/tools/winui-analyzer/Microsoft.WindowsAppSDK.Analyzers.Tests/` | payloads-and-tests |
 | `src/tools/winui-analyzer/RULES.md` / `CHANGELOG.md` | docs-and-manifests |
-| `src/tools/winmd-cli/`, `src/tools/winui-search/` | tool-correctness |
+| `src/tools/winmd-cli/` | tool-correctness |
 | `scripts/build-tools.ps1` | payloads-and-tests |
 | `.github/workflows/` | docs-and-manifests (CI), payloads-and-tests (provenance) |
 | `README.md`, `SECURITY.md`, `SUPPORT.md` | docs-and-manifests |
@@ -234,7 +234,7 @@ verdict.
 - **No file output.** Stdout only, unless the user explicitly asked for a
   file.
 - **No build/test execution.** Flag staleness (analyzer DLL not refreshed,
-  `winui-search.exe` not refreshed, `RULES.md` not updated) but do not run
+  `RULES.md` not updated) but do not run
   `scripts/build-tools.ps1` or `dotnet test` yourself — they are slow and
   the contributor will run them.
 - **Signal-to-noise.** Reject sub-agent findings that are pure style nits,

.github/skills/pr-review/dimensions/_shared-contract.md

@@ -98,7 +98,7 @@ high. The bar for adding tooling enforcement is lower.
   is too scenario-specific.
 - Trigger-phrasing problems in `description:` frontmatter that would
   cause the wrong agent activation.
-- Stale committed payloads (analyzer DLL, `winui-search.exe`,
+- Stale committed analyzer payloads (DLL or
   `Microsoft.WindowsAppSDK.Analyzers.targets`) that will fail CI
   provenance.
 - Manifest / version / agent-file drift that ships broken artifacts to
@@ -136,8 +136,8 @@ change should land. Cite the tier on every `skill-content` and
 
 | Tier | Type | Reliability | Examples in this repo |
 |------|------|-------------|------------------------|
-| **0** | Environment / harness defaults | Highest — agent never sees it | `dotnet new` template choice, `BuildAndRun.ps1` defaults, prerequisite checks in `winui-setup` |
-| **1** | Tooling enforcement | High — produces errors/warnings the agent must address | `Microsoft.WindowsAppSDK.Analyzers` rules, `winui-search.exe` query results, `winmd.exe` API verification, `winapp` CLI exit codes |
+| **0** | Environment / harness defaults | Highest — agent never sees it | `winapp new` template choice, `BuildAndRun.ps1` defaults, prerequisite checks in `winui-setup` |
+| **1** | Tooling enforcement | High — produces errors/warnings the agent must address | `Microsoft.WindowsAppSDK.Analyzers` rules, `winapp find-ui` query results, `winmd.exe` API verification, `winapp` CLI exit codes |
 | **2** | Templates / scaffolding | Medium — structural, applied at creation time | `Microsoft.WindowsAppSDK.WinUI.CSharp.Templates`, starter project files |
 | **3** | Instructions / skills | Lowest — advisory, frequently ignored | `SKILL.md` content, `winui-dev.agent.md` rules, `references/*.md` |
 

.github/skills/pr-review/dimensions/docs-and-manifests.md

@@ -21,13 +21,14 @@ When code or skills change, these need to keep up:
 - `plugins/winui/{.claude-plugin,.codex-plugin}/plugin.json` and
   `openclaw.plugin.json` — legacy-client adapters pointing to the canonical
   `agent-plugin/skills/` directory.
-- `plugins/winui/agents/winui-dev.agent.md` — the orchestrator agent
-  prompt; mentions specific skills by name and lists default-loaded
-  skills.
+- `plugins/winui/agents/winui-dev.agent.md` and
+  `plugins/winui/agent-plugin/com.github.copilot/agents/winui-dev.agent.md` —
+  compatibility and Copilot orchestrator prompts; mention specific skills by
+  name and list default-loaded skills.
 - `src/tools/winui-analyzer/RULES.md` — rule catalog (per-rule entry
   required for every shipped diagnostic; IDs are immutable).
 - `src/tools/winui-analyzer/CHANGELOG.md` — analyzer-scoped changelog.
-- Per-tool READMEs: `src/tools/{winui-analyzer,winmd-cli,winui-search}/README.md`.
+- Per-tool READMEs: `src/tools/{winui-analyzer,winmd-cli}/README.md`.
 - `SECURITY.md`, `SUPPORT.md`, `THIRD_PARTY_NOTICES.md`,
   `cgmanifest.json` — only relevant when dependencies or contact
   surfaces change.
@@ -40,10 +41,10 @@ When code or skills change, these need to keep up:
 
 - **New skill added under `plugins/winui/agent-plugin/skills/<new>/`** without a
   matching row in `README.md`'s "eight skills" table → **high**.
-- **Skill renamed.** `plugins/winui/agents/winui-dev.agent.md`
-  references skills by name (e.g. "Load the `winui-dev-workflow`
-  skill"). Renames must update every mention in the agent file
-  *and* in any sibling skill that links to it. → **high**.
+- **Skill renamed.** Both orchestrator agent files reference skills by name
+  (e.g. "Load the `winui-dev-workflow` skill"). Renames must update every
+  mention in both agent files and in any sibling skill that links to it.
+  → **high**.
 - **Skill removed without README update.** Same as above, inverse.
 - **Skill description copy doesn't match `description:` frontmatter.**
   README's table is hand-curated; the canonical text lives in the
@@ -102,7 +103,7 @@ When code or skills change, these need to keep up:
   walks `find plugins/winui/agent-plugin/skills -type f -name SKILL.md`. New
   skills outside this glob won't be validated → **medium**.
 - Any CI step's hardcoded file path
-  (e.g. `plugins/winui/agent-plugin/skills/winui-design/winui-search.exe`)
+  (e.g. `plugins/winui/agent-plugin/skills/winui-dev-workflow/analyzer/Microsoft.WindowsAppSDK.Analyzers.dll`)
   changed in the diff but not in the workflow → **high**.
 
 ### Other docs

.github/skills/pr-review/dimensions/payloads-and-tests.md

@@ -6,21 +6,20 @@ Set `Domain: payloads-and-tests` on every finding.
 
 ## What this dimension owns
 
-This repo commits **prebuilt binary payloads** alongside source for
-two of its tools, and the CI `pr-validation` workflow has provenance
-jobs that fail the PR if those payloads drift from source. This
-dimension's job is to flag drift *before* the contributor pushes.
+This repo commits the analyzer's **prebuilt payloads** alongside source,
+and the CI `pr-validation` workflow has provenance jobs that fail the PR
+if those payloads drift from source. This dimension's job is to flag
+drift *before* the contributor pushes.
 
 The payloads:
 
 | Payload (committed) | Source | CI job that catches drift |
 |---|---|---|
 | `plugins/winui/agent-plugin/skills/winui-dev-workflow/analyzer/Microsoft.WindowsAppSDK.Analyzers.dll` | `src/tools/winui-analyzer/Microsoft.WindowsAppSDK.Analyzers/` | `analyzer-provenance` (sha256 + size delta) |
 | `plugins/winui/agent-plugin/skills/winui-dev-workflow/analyzer/Microsoft.WindowsAppSDK.Analyzers.targets` | `src/tools/winui-analyzer/Microsoft.WindowsAppSDK.Analyzers/Microsoft.WindowsAppSDK.Analyzers.targets` | `analyzer-targets-sync` (byte-identical) |
-| `plugins/winui/agent-plugin/skills/winui-design/winui-search.exe` | `src/tools/winui-search/` | `winui-search-provenance` (smoke + size ±10%) |
 
-Refresh command: `./scripts/build-tools.ps1` (no flags) rebuilds all
-three tools and refreshes both payloads. The contributor will run
+Refresh command: `./scripts/build-tools.ps1` (no flags) rebuilds both
+C# tools and refreshes the analyzer payloads. The contributor will run
 this — you only flag drift.
 
 ## What to look for
@@ -39,21 +38,11 @@ this — you only flag drift.
   Same pattern but for
   `Microsoft.WindowsAppSDK.Analyzers.targets`. CI
   `analyzer-targets-sync` requires byte-identical files. → **high**.
-- **`winui-search` source touched, exe not refreshed.** Diff
-  includes `src/tools/winui-search/**` *but not*
-  `plugins/winui/agent-plugin/skills/winui-design/winui-search.exe` → **high**.
-  CI `winui-search-provenance` runs smoke tests (`list`, `search
-  tabview` → must include `gallery-tabview`) and a ±10% size
-  comparison.
 - **Payload-only change (no source).** The inverse — committed DLL
-  or exe updated without a corresponding source diff. Either the
+  or `.targets` updated without a corresponding source diff. Either the
   source was already on `main` (fine — this is a refresh PR) or
   someone hand-edited the binary (red flag). → **medium**, ask the
   contributor to confirm.
-- **`winui-search` data files updated without rebuild.** Diff
-  touches `src/tools/winui-search/Data/*.json` *but not* the
-  committed exe. The data is **embedded** at build time — JSON
-  changes have no effect until the exe is republished. → **high**.
 
 ### Analyzer test coverage
 
@@ -124,10 +113,8 @@ findings, but additionally:
 
 ## Severity guide for this dimension
 
-- Source touched but committed payload not refreshed (DLL, .targets,
-  or exe) → **high** (CI provenance will fail).
-- `winui-search` data JSON edited without exe rebuild → **high**
-  (the change is invisible at runtime).
+- Source touched but committed payload not refreshed (DLL or .targets)
+  → **high** (CI provenance will fail).
 - New analyzer rule with no test → **high**.
 - New tool not wired into `build-tools.ps1` → **medium**.
 - Allowlist change without regression test → **medium**.

.github/skills/pr-review/dimensions/skill-content.md

@@ -14,8 +14,9 @@ This dimension reviews **prose changes** to:
 - `plugins/winui/agent-plugin/skills/<name>/SKILL.md` (the 8 shipped skills)
 - `plugins/winui/agent-plugin/skills/<name>/references/*.md` (deep-dive references
   loaded on demand)
-- `plugins/winui/agents/winui-dev.agent.md` (the orchestrator agent
-  prompt)
+- `plugins/winui/agents/winui-dev.agent.md` and
+  `plugins/winui/agent-plugin/com.github.copilot/agents/winui-dev.agent.md`
+  (orchestrator agent prompts)
 - `.github/skills/<name>/SKILL.md` (repo-internal skills like this one)
 
 The audience for this content is **AI coding agents** (Copilot CLI,
@@ -62,7 +63,7 @@ priors. Every line of prose must pull its weight.
   re-state rules already covered in a sibling skill — link or
   reference instead. Flag when the same checklist appears twice.
 - **Restating help text.** Bullets that re-document what `winapp
-  --help`, `winui-search list`, or `winmd --help` already prints are
+  --help`, `winapp find-ui --help`, or `winmd --help` already prints are
   pure context inflation.
 - **"Best practices" laundry lists** with no concrete change in agent
   behavior (e.g. "follow YAGNI / DRY / KISS") — these are universally
@@ -89,8 +90,8 @@ priors. Every line of prose must pull its weight.
   read.
 - **References that should be tool data.** A long reference file that
   is essentially a catalogue of WinUI controls, snippets, or API
-  signatures duplicates `winui-search.exe`'s data. Flag and emit a
-  paired `skill-tool-boundary` finding (Tier 1 alternative).
+  signatures duplicates `winapp find-ui` or `winmd` data. Flag and emit
+  a paired `skill-tool-boundary` finding (Tier 1 alternative).
 
 ### Trigger / activation hygiene
 

.github/skills/pr-review/dimensions/skill-tool-boundary.md

@@ -20,8 +20,8 @@ consider them before defaulting to Tier 3 prose.
 
 | Tier | Type | Reliability | In-repo examples |
 |------|------|-------------|-------------------|
-| **0** | Environment / harness defaults | Highest — agent never sees it | `dotnet new` template choice baked into `winui-dev-workflow`, `BuildAndRun.ps1` defaults, `winui-setup` prerequisite checks |
-| **1** | Tooling enforcement | High — produces diagnostics agent must address | `Microsoft.WindowsAppSDK.Analyzers` rules (WUI0xxx-WUI4xxx), `winui-search.exe` queries, `winmd.exe` API verification, `winapp` CLI exit codes |
+| **0** | Environment / harness defaults | Highest — agent never sees it | `winapp new` template choice, `BuildAndRun.ps1` defaults, `winui-setup` prerequisite checks |
+| **1** | Tooling enforcement | High — produces diagnostics agent must address | `Microsoft.WindowsAppSDK.Analyzers` rules (WUI0xxx-WUI4xxx), `winapp find-ui` queries, `winmd.exe` API verification, `winapp` CLI exit codes |
 | **2** | Templates / scaffolding | Medium — structural, applied once | `Microsoft.WindowsAppSDK.WinUI.CSharp.Templates`, starter `.csproj` defaults |
 | **3** | Instructions / skills | Lowest — advisory, frequently ignored | `SKILL.md` content, `winui-dev.agent.md` rules, `references/*.md` |
 
@@ -53,19 +53,22 @@ This is the most common drift. Symptoms:
   This is what `winmd.exe` is for — recommend invoking it from the
   skill instead of duplicating the rule.
 - A new list of **WinUI control names or sample patterns**. This is
-  what `winui-search.exe` is for. The skill should *describe how to
+  what `winapp find-ui` is for. The skill should *describe how to
   query* the tool, not embed the catalogue.
-- A new bullet that says **"after building, do X"**. This usually
-  belongs in `BuildAndRun.ps1` (Tier 0) so the agent gets it for free.
+- A new bullet that says **"after building, do X"**. General build/run
+  behavior belongs in the upstream `winapp` command so every caller gets it;
+  reserve `BuildAndRun.ps1` for analyzer injection and diagnostic defaults.
 - A new "common error" entry that boils down to a missing prerequisite
   — that belongs in `winui-setup` (Tier 0/1).
 
 For each such finding:
 - Severity: **medium** (would be high if the analyzer/tool clearly
   could enforce it but the contributor chose prose).
 - Recommendation: name the specific Tier 1 hook — "Add an analyzer
-  rule under `WUI20xx` (runtime/layout/XAML pitfalls)", "Extend
-  `winui-search` index", "Add a step to `BuildAndRun.ps1`".
+  rule under `WUI20xx` (runtime/layout/XAML pitfalls)", "Improve the
+  upstream `winapp find-ui` corpus", or "Improve the relevant upstream
+  `winapp` subcommand". Recommend `BuildAndRun.ps1` only for analyzer or
+  diagnostic-default behavior.
 
 ### Skill prose that should be a Tier 2 (template) change
 
@@ -139,9 +142,6 @@ Less common but real:
   than a real WinUI pitfall. Analyzer noise erodes trust in the
   whole catalog. Recommend converting to skill guidance (Tier 3) or
   dropping.
-- A new `winui-search` synonym hack that encodes a single
-  contributor's mental model rather than a generally useful query
-  alias.
 - A new `winmd-cli` flag that exists only to match one skill's exact
   output format — couplings like this should be inverted (skill
   adapts to tool, not vice versa).
@@ -179,8 +179,9 @@ justification, the addition is misplaced — emit a finding.
 
 - New skill prose duplicating an existing analyzer rule → **medium**
   (Tier 3, recommend cite-the-rule-instead).
-- New skill prose that should clearly have been a new analyzer rule
-  / new `winui-search` data / new `BuildAndRun.ps1` step → **high**
+- New skill prose that should clearly have been a new analyzer rule,
+  an upstream `winapp find-ui` improvement, or a new `BuildAndRun.ps1`
+  step → **high**
   if the change is large and the Tier 1 path is straightforward;
   **medium** otherwise.
 - New skill prose that should clearly have been a `winapp` CLI