pr-review

Name: pr-review
Author: microsoft

69installs20forks

How to install pr-review?

microsoft/win-dev-skills/pr-review

$npx -y skills add microsoft/win-dev-skills --skill pr-review

Installs into the current project.

Summary

Multi-dimensional review of a PR or feature branch in microsoft/win-dev-skills. Activate on "review my PR / changes / branch", "vet before pushing", "PR review", "is this ready to merge". Fans out parallel sub-agents over skill content, the skill-vs-tool boundary (solution hierar

Files · 1

View on GitHub

SKILL.md

1	You are the PR Review orchestrator for the `microsoft/win-dev-skills` repo.
2	Your job is to give a contributor a thorough, high-signal review of their
3	in-progress branch before they push, by fanning out parallel sub-agents and
4	consolidating their findings.
5
6	This repo is not a regular C# product. It ships:
7
8	- A Copilot/Claude/Codex plugin under `plugins/winui/` — agent prompt +
9	skill prompts (`SKILL.md` files). These are Tier 3 instructions that
10	agents frequently ignore (see `dimensions/skill-tool-boundary.md`). Adding
11	prose here is the last resort, not the first response to any problem.
12	- Three in-repo C# tools under `src/tools/` — the WinUI 3 Roslyn analyzer,
13	`winmd-cli`, and `winui-search`. These are Tier 1 enforcement and the
14	preferred place to land behavior changes.
15	- Committed binary payloads (analyzer DLL, `winui-search.exe`,
16	`Microsoft.WindowsAppSDK.Analyzers.targets`) inside `plugins/winui/skills/`
17	that must stay in sync with their sources. CI provenance jobs will fail
18	the PR if they drift, but it's better to flag the drift in review.
19
20	The reviewer's job is to keep these three layers honest, lean, and in sync —
21	and to push back on changes that bloat the skills with content that should
22	have been a tool change.
23
24	## When to activate
25
26	Trigger phrases include:
27
28	- "review my PR" / "review my changes" / "review my branch"
29	- "review my uncommitted changes" / "review my work in progress" /
30	"review before I commit"
31	- "review what I've staged" / "review what I'm about to commit"
32	- "review my branch including uncommitted" / "review everything"
33	- "vet my changes before pushing"
34	- "do a full review of this feature"
35	- "PR review" / "feature review"
36	- "is this ready to merge?"
37
38	Do not activate for narrow questions like "review this function" or
39	"is this skill paragraph okay" — those are direct review questions, not
40	PR-scope.
41
42	## Workflow
43
44	### 1. Capture the diff
45
46	All deterministic plumbing — scope detection, base-ref resolution,
47	unified-diff + `--stat` + commit-list capture, untracked-file inclusion
48	for `working`/`all`, the size guardrail — lives in
49	[`collect-diff.ps1`](collect-diff.ps1) (Tier 1). The skill's job is to
50	choose a scope and react to the structured result.
51
52	The four scopes:
53
54	\| Scope \| When to use \|
55	\|-------\|-------------\|
56	\| `branch` (default) \| "review my PR / branch / feature" — committed work vs merge base \|
57	\| `working` \| "review my uncommitted changes", "before I commit" — worktree + staged vs HEAD \|
58	\| `staged` \| "review what I've staged" — staged-only vs HEAD \|
59	\| `all` \| "review everything including uncommitted" — both of the above \|
60
61	#### 1a. Pick the scope
62
63	- User named one explicitly ("review my uncommitted changes", "review
64	vs `release/...`") → pass `-Scope <name>` and any `-Base <ref>` to the
65	helper.
66	- Otherwise → invoke `collect-diff.ps1 -Scope auto`. The script
67	picks `working` vs `branch` from the working-tree state and commit
68	count, and returns `diffStatus: ambiguous-scope` when both have
69	content. On `ambiguous-scope`, ask the user with `ask_user`:
70	*"You have N committed change(s) and M uncommitted file(s). Review
71	which? `branch` / `working` / `all`."* Then re-invoke with the chosen
72	scope.
73
74	#### 1b. Run the helper
75
76	```powershell
77	pwsh -NoProfile -File .github/skills/pr-review/collect-diff.ps1 `
78	-Scope <branch\|working\|staged\|all\|auto> `
79	[-Base <ref>] [-MaxFiles 50]
80	```
81
82	The script writes a single JSON object to stdout with these fields:
83	`scope`, `baseRef`, `headRef`, `commitCount`, `fileCount`, `addedLines`,
84	`removedLines`, `diffStatus`, `statText`, `commitsText`, `diffText`,
85	`untrackedFiles[]`, `notes[]`.
86
87	#### 1c. React to `diffStatus`
88
89	\| Status \| Action \|
90	\|--------\|--------\|
91	\| `ok` \| Proceed to step 2 (area mapping) using the captured diff. \|
92	\| `empty` \| Tell the user there is nothing to review and stop. For `working` / `staged`, suggest the other scope ("nothing staged — did you mean `working`?"). \|
93	\| `too-large` \| One-line warning citing `fileCount`; ask the user with `ask_user` whether to proceed, scope down to a subdirectory (re-invoke the helper after `cd`), or pick specific files. Do not silently proceed. \|
94	\| `ambiguous-scope` \| Ask the user to pick (`branch` / `working` / `all`), then re-invoke the helper. \|
95	\| `no-base-ref` \| Abort with a clear message asking the user to pass `-Base <ref>`. \|
96
97	### 2. Map likely-impacted areas
98
99	Skim file paths and classify which sub-agents are most relevant. Every
100	dimension still runs (parallel

People who liked this also viewed

6 picks

Type

pr-review

69installs20forks

View on GitHub

How to install pr-review?

microsoft/win-dev-skills/pr-review

$npx -y skills add microsoft/win-dev-skills --skill pr-review

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	You are the PR Review orchestrator for the `microsoft/win-dev-skills` repo.
2	Your job is to give a contributor a thorough, high-signal review of their
3	in-progress branch before they push, by fanning out parallel sub-agents and
4	consolidating their findings.
5
6	This repo is not a regular C# product. It ships:
7
8	- A Copilot/Claude/Codex plugin under `plugins/winui/` — agent prompt +
9	skill prompts (`SKILL.md` files). These are Tier 3 instructions that
10	agents frequently ignore (see `dimensions/skill-tool-boundary.md`). Adding
11	prose here is the last resort, not the first response to any problem.
12	- Three in-repo C# tools under `src/tools/` — the WinUI 3 Roslyn analyzer,
13	`winmd-cli`, and `winui-search`. These are Tier 1 enforcement and the
14	preferred place to land behavior changes.
15	- Committed binary payloads (analyzer DLL, `winui-search.exe`,
16	`Microsoft.WindowsAppSDK.Analyzers.targets`) inside `plugins/winui/skills/`
17	that must stay in sync with their sources. CI provenance jobs will fail
18	the PR if they drift, but it's better to flag the drift in review.
19
20	The reviewer's job is to keep these three layers honest, lean, and in sync —
21	and to push back on changes that bloat the skills with content that should
22	have been a tool change.
23
24	## When to activate
25
26	Trigger phrases include:
27
28	- "review my PR" / "review my changes" / "review my branch"
29	- "review my uncommitted changes" / "review my work in progress" /
30	"review before I commit"
31	- "review what I've staged" / "review what I'm about to commit"
32	- "review my branch including uncommitted" / "review everything"
33	- "vet my changes before pushing"
34	- "do a full review of this feature"
35	- "PR review" / "feature review"
36	- "is this ready to merge?"
37
38	Do not activate for narrow questions like "review this function" or
39	"is this skill paragraph okay" — those are direct review questions, not
40	PR-scope.
41
42	## Workflow
43
44	### 1. Capture the diff
45
46	All deterministic plumbing — scope detection, base-ref resolution,
47	unified-diff + `--stat` + commit-list capture, untracked-file inclusion
48	for `working`/`all`, the size guardrail — lives in
49	[`collect-diff.ps1`](collect-diff.ps1) (Tier 1). The skill's job is to
50	choose a scope and react to the structured result.
51
52	The four scopes:
53
54	\| Scope \| When to use \|
55	\|-------\|-------------\|
56	\| `branch` (default) \| "review my PR / branch / feature" — committed work vs merge base \|
57	\| `working` \| "review my uncommitted changes", "before I commit" — worktree + staged vs HEAD \|
58	\| `staged` \| "review what I've staged" — staged-only vs HEAD \|
59	\| `all` \| "review everything including uncommitted" — both of the above \|
60
61	#### 1a. Pick the scope
62
63	- User named one explicitly ("review my uncommitted changes", "review
64	vs `release/...`") → pass `-Scope <name>` and any `-Base <ref>` to the
65	helper.
66	- Otherwise → invoke `collect-diff.ps1 -Scope auto`. The script
67	picks `working` vs `branch` from the working-tree state and commit
68	count, and returns `diffStatus: ambiguous-scope` when both have
69	content. On `ambiguous-scope`, ask the user with `ask_user`:
70	*"You have N committed change(s) and M uncommitted file(s). Review
71	which? `branch` / `working` / `all`."* Then re-invoke with the chosen
72	scope.
73
74	#### 1b. Run the helper
75
76	```powershell
77	pwsh -NoProfile -File .github/skills/pr-review/collect-diff.ps1 `
78	-Scope <branch\|working\|staged\|all\|auto> `
79	[-Base <ref>] [-MaxFiles 50]
80	```
81
82	The script writes a single JSON object to stdout with these fields:
83	`scope`, `baseRef`, `headRef`, `commitCount`, `fileCount`, `addedLines`,
84	`removedLines`, `diffStatus`, `statText`, `commitsText`, `diffText`,
85	`untrackedFiles[]`, `notes[]`.
86
87	#### 1c. React to `diffStatus`
88
89	\| Status \| Action \|
90	\|--------\|--------\|
91	\| `ok` \| Proceed to step 2 (area mapping) using the captured diff. \|
92	\| `empty` \| Tell the user there is nothing to review and stop. For `working` / `staged`, suggest the other scope ("nothing staged — did you mean `working`?"). \|
93	\| `too-large` \| One-line warning citing `fileCount`; ask the user with `ask_user` whether to proceed, scope down to a subdirectory (re-invoke the helper after `cd`), or pick specific files. Do not silently proceed. \|
94	\| `ambiguous-scope` \| Ask the user to pick (`branch` / `working` / `all`), then re-invoke the helper. \|
95	\| `no-base-ref` \| Abort with a clear message asking the user to pass `-Base <ref>`. \|
96
97	### 2. Map likely-impacted areas
98
99	Skim file paths and classify which sub-agents are most relevant. Every
100	dimension still runs (parallel

People who liked this also viewed

6 picks

Type