qt-qml-review

Name: qt-qml-review
Author: TheQtCompanyRnD

384installs33forks

How to install qt-qml-review?

theqtcompanyrnd/agent-skills/qt-qml-review

$npx -y skills add TheQtCompanyRnD/agent-skills --skill qt-qml-review

Installs into the current project.

Summary

Invoke when the user asks to review, check, audit, or look over Qt6 QML code -- or suggest before committing. Runs deterministic linting (47+ rules) then six parallel deep- analysis agents covering bindings, layout, loaders, delegates, states, and performance. Optionally invokes

Files · 1

View on GitHub

SKILL.md

1	# Qt QML Code Review
2
3	A structured, read-only code review skill for Qt6 QML code that
4	combines deterministic linting with parallel agent-driven deep
5	analysis across six focused domains.
6
7	## When to use this skill
8
9	- When the user mentions review-related tasks: "review", "check",
10	"audit", "look over", "code review", "sanity check"
11	- Suggest running this skill before committing QML code
12	- When the user asks to validate Qt6 QML code quality
13
14	## Scope detection
15
16	Detect the user's intended scope from their language:
17
18	### Diff/commit scope (narrow)
19	Triggered by language like: "this commit", "these changes",
20	"the diff", "what I changed", "my changes", "staged changes",
21	"outstanding changes", "before I commit"
22
23	Action: Run `git diff` (unstaged) and `git diff --cached`
24	(staged) to obtain the changeset. If the user says "this commit",
25	use `git diff HEAD~1..HEAD`. Review only the changed lines plus
26	sufficient surrounding context (±50 lines) for understanding.
27	Only report issues found in the changed lines -- do not report
28	issues in unchanged surrounding context.
29
30	### Codebase scope (wide)
31	Triggered by language like: "review the codebase", "audit the
32	project", "check the repository", "review src/", or when a specific
33	file/directory path is given without commit language.
34
35	Action: Glob for `*.qml` files in the specified scope. Review
36	all matched files.
37
38	## Execution order
39
40	The review proceeds in three phases. Never skip a phase.
41
42	### Phase 1: Deterministic linting (Python script)
43
44	Run the unified Python linter against the target files. Requires
45	Python 3.6+ (no external dependencies). If Python is not available,
46	warn the user and skip to Phase 1b.
47
48	```bash
49	python3 references/lint-scripts/qt_qml_lint.py <files...>
50	# If python3 is not found, fall back to:
51	python references/lint-scripts/qt_qml_lint.py <files...>
52	```
53
54	This single-pass scanner encodes all mechanically-checkable rules
55	from the QML review checklist. It reads each file once and evaluates
56	all rules per line, plus block-level structural checks. Output is
57	deterministic and repeatable. The linter is authoritative -- do not
58	second-guess its output.
59
60	Collect all output before proceeding.
61
62	Rule categories (47+ checks):
63	- IMP (Imports) -- ordering, versioning, redundancy, deprecation
64	- ORD (Ordering) -- QML attribute ordering convention
65	- BND (Bindings) -- property var, imperative =, Qt.binding style
66	- LAY (Layout) -- anchors/Layout mixing, sizing in layouts
67	- LDR (Loader) -- status guards, createComponent, createQmlObject
68	- DEL (Delegates) -- required properties, reuse safety, connect()
69	- STA (States) -- PropertyChanges syntax, transitions, StateGroup
70	- IMG (Images) -- sourceSize, asynchronous loading
71	- PRF (Performance) -- transparent rect, opacity, clip, layer
72	- STY (Style) -- id:root, camelCase, group notation
73	- SIG (Signals) -- Connections target, handler syntax
74	- ERR (Error/Security) -- hardcoded http://, non-portable paths
75	- JS (JavaScript) -- var/let/const, loose equality
76
77	### Phase 1b: System qmllint (optional)
78
79	Attempt to run `qmllint` if available on the system. Detection
80	order:
81
82	1. `$QT_HOST_PATH/bin/qmllint`
83	2. `which qmllint` / `where qmllint`
84	3. Skip if not found (warn user)
85
86	If found, run with JSON output:
87
88	```bash
89	qmllint --json - -I <import-paths> <files...>
90	```
91
92	Parse the JSON output and merge with Python linter findings.
93	Deduplicate by file+line+issue. qmllint is authoritative for type-
94	level checks (unresolved types, incompatible assignments, alias
95	cycles). The Python linter is authoritative for style, ordering,
96	and performance patterns that qmllint does not cover.
97
98	### Phase 2: Agent-driven deep analysis (6 parallel agents)
99
100	Launch six focused review agents in parallel. Name each agent
101	descriptively when launching (e.g. "Agent 1: Bindings & Properties")
102	to provide progress visibility. Each agent has a tight scope
103	and a specific checklist. Agents are READ-ONLY -- they must
104	never edit or write files.
105
106	Tool-agnostic agent contract: Each agent described below is
107	a self-contained review mission. In Claude Code, launch them as
108	general-purpose subagents. In other tools, implement each as
109	whatever subprocess, prompt chain, or analysis pass the tool
110	supports. T

People who liked this also viewed

6 picks

Type

qt-qml-review

384installs33forks

View on GitHub

How to install qt-qml-review?

theqtcompanyrnd/agent-skills/qt-qml-review

$npx -y skills add TheQtCompanyRnD/agent-skills --skill qt-qml-review

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Qt QML Code Review
2
3	A structured, read-only code review skill for Qt6 QML code that
4	combines deterministic linting with parallel agent-driven deep
5	analysis across six focused domains.
6
7	## When to use this skill
8
9	- When the user mentions review-related tasks: "review", "check",
10	"audit", "look over", "code review", "sanity check"
11	- Suggest running this skill before committing QML code
12	- When the user asks to validate Qt6 QML code quality
13
14	## Scope detection
15
16	Detect the user's intended scope from their language:
17
18	### Diff/commit scope (narrow)
19	Triggered by language like: "this commit", "these changes",
20	"the diff", "what I changed", "my changes", "staged changes",
21	"outstanding changes", "before I commit"
22
23	Action: Run `git diff` (unstaged) and `git diff --cached`
24	(staged) to obtain the changeset. If the user says "this commit",
25	use `git diff HEAD~1..HEAD`. Review only the changed lines plus
26	sufficient surrounding context (±50 lines) for understanding.
27	Only report issues found in the changed lines -- do not report
28	issues in unchanged surrounding context.
29
30	### Codebase scope (wide)
31	Triggered by language like: "review the codebase", "audit the
32	project", "check the repository", "review src/", or when a specific
33	file/directory path is given without commit language.
34
35	Action: Glob for `*.qml` files in the specified scope. Review
36	all matched files.
37
38	## Execution order
39
40	The review proceeds in three phases. Never skip a phase.
41
42	### Phase 1: Deterministic linting (Python script)
43
44	Run the unified Python linter against the target files. Requires
45	Python 3.6+ (no external dependencies). If Python is not available,
46	warn the user and skip to Phase 1b.
47
48	```bash
49	python3 references/lint-scripts/qt_qml_lint.py <files...>
50	# If python3 is not found, fall back to:
51	python references/lint-scripts/qt_qml_lint.py <files...>
52	```
53
54	This single-pass scanner encodes all mechanically-checkable rules
55	from the QML review checklist. It reads each file once and evaluates
56	all rules per line, plus block-level structural checks. Output is
57	deterministic and repeatable. The linter is authoritative -- do not
58	second-guess its output.
59
60	Collect all output before proceeding.
61
62	Rule categories (47+ checks):
63	- IMP (Imports) -- ordering, versioning, redundancy, deprecation
64	- ORD (Ordering) -- QML attribute ordering convention
65	- BND (Bindings) -- property var, imperative =, Qt.binding style
66	- LAY (Layout) -- anchors/Layout mixing, sizing in layouts
67	- LDR (Loader) -- status guards, createComponent, createQmlObject
68	- DEL (Delegates) -- required properties, reuse safety, connect()
69	- STA (States) -- PropertyChanges syntax, transitions, StateGroup
70	- IMG (Images) -- sourceSize, asynchronous loading
71	- PRF (Performance) -- transparent rect, opacity, clip, layer
72	- STY (Style) -- id:root, camelCase, group notation
73	- SIG (Signals) -- Connections target, handler syntax
74	- ERR (Error/Security) -- hardcoded http://, non-portable paths
75	- JS (JavaScript) -- var/let/const, loose equality
76
77	### Phase 1b: System qmllint (optional)
78
79	Attempt to run `qmllint` if available on the system. Detection
80	order:
81
82	1. `$QT_HOST_PATH/bin/qmllint`
83	2. `which qmllint` / `where qmllint`
84	3. Skip if not found (warn user)
85
86	If found, run with JSON output:
87
88	```bash
89	qmllint --json - -I <import-paths> <files...>
90	```
91
92	Parse the JSON output and merge with Python linter findings.
93	Deduplicate by file+line+issue. qmllint is authoritative for type-
94	level checks (unresolved types, incompatible assignments, alias
95	cycles). The Python linter is authoritative for style, ordering,
96	and performance patterns that qmllint does not cover.
97
98	### Phase 2: Agent-driven deep analysis (6 parallel agents)
99
100	Launch six focused review agents in parallel. Name each agent
101	descriptively when launching (e.g. "Agent 1: Bindings & Properties")
102	to provide progress visibility. Each agent has a tight scope
103	and a specific checklist. Agents are READ-ONLY -- they must
104	never edit or write files.
105
106	Tool-agnostic agent contract: Each agent described below is
107	a self-contained review mission. In Claude Code, launch them as
108	general-purpose subagents. In other tools, implement each as
109	whatever subprocess, prompt chain, or analysis pass the tool
110	supports. T

People who liked this also viewed

6 picks

Type