ios-memgraph-analysis

Name: ios-memgraph-analysis
Author: dpearson2699

158installs45forks

How to install ios-memgraph-analysis?

dpearson2699/swift-ios-skills/ios-memgraph-analysis

$npx -y skills add dpearson2699/swift-ios-skills --skill ios-memgraph-analysis

Installs into the current project.

Summary

Use when capturing or analyzing an iOS .memgraph, especially when the task mentions a memory leak, heap growth, persistent memory increase, ownership path, or matched-capture comparison with Apple CLI tools. Covers unambiguous Simulator capture, leaks/heap/vmmap/malloc_history ev

Files · 1

View on GitHub

SKILL.md

1	# iOS Memgraph Analysis
2
3	Use memory graphs to prove why memory survives a defined lifetime boundary.
4	Separate unreachable leaks from reachable growth, preserve raw tool output, and
5	verify the same app-owned type and ownership path after a fix.
6
7	## Contents
8
9	- [Boundary](#boundary)
10	- [Evidence Model](#evidence-model)
11	- [Workflow](#workflow)
12	- [Ownership Decisions](#ownership-decisions)
13	- [Common Mistakes](#common-mistakes)
14	- [Review Checklist](#review-checklist)
15	- [References](#references)
16
17	## Boundary
18
19	This skill owns `.memgraph` capture and command-line ownership/growth analysis.
20	Use the Memory Graph Debugger or Instruments when their interactive graph and
21	allocation timeline are the primary task. Use source review for a suspected
22	closure capture only after runtime evidence identifies the lifetime or path.
23
24	## Evidence Model
25
26	Do not collapse these conditions:
27
28	- Unreachable leak: allocated memory no longer has a path from a live root.
29	An isolated strong cycle can be unreachable and still consume memory.
30	- Reachable but abandoned state: a live root still retains objects the user
31	flow no longer needs. `leaks` may correctly report zero.
32	- Expected cache or pool: memory survives intentionally and must be judged by
33	its bound, eviction behavior, and pressure response.
34	- Heap regression or fragmentation: footprint grows because more/larger
35	allocations persist or dirty pages are poorly utilized, without a leak.
36
37	Apple's leak scanner uses conservative pointer discovery and incomplete type
38	metadata. Counts can fluctuate, and a zero result does not prove the absence of
39	an ownership bug. Strong evidence identifies the expected lifetime, an
40	app-owned type or allocation, and a credible path or isolated reproduction.
41
42	## Workflow
43
44	### 1. Define the lifetime before capturing
45
46	Name the object that should disappear and the event that ends its useful life.
47	For example: `EditorViewModel` should deinitialize after dismissing the editor
48	and completing pending save work.
49
50	Record one deterministic sequence:
51
52	1. launch or restore a known state;
53	2. take an optional baseline graph;
54	3. perform the feature flow;
55	4. cross the expected release boundary;
56	5. wait for legitimate asynchronous cleanup;
57	6. take the post-flow graph.
58
59	Keep build, simulator/device, data, Malloc Stack Logging setting, and repetition
60	count stable. Malloc Stack Logging adds valuable allocation backtraces but also
61	overhead; compare only runs with the same setting.
62
63	### 2. Capture a graph without guessing the process
64
65	Xcode can export a graph from the Memory Graph Debugger. For a running Simulator
66	app, use the helper from this skill:
67
68	```bash
69	mkdir -p /tmp/myapp-memory
70	mkdir /tmp/myapp-memory/run-01
71	python3 scripts/capture_sim_memgraph.py \
72	--bundle-id com.example.MyApp \
73	--output-dir /tmp/myapp-memory/run-01 \
74	--pretty > /tmp/myapp-memory/run-01/capture.json
75	```
76
77	The per-run `mkdir` must fail if the capture directory already exists. Use a
78	new run name rather than mixing stale evidence with a retry.
79
80	Pass `--udid` when more than one Simulator is booted. The helper accepts only
81	one exact launchd label and PID; zero or multiple matches are errors. It runs the
82	host `leaks --outputGraph` command, retains stdout/stderr, and writes a manifest.
83	Do not replace this with `pgrep \| head -1` or a substring match.
84
85	Capturing suspends the process. Do not use capture latency as performance data.
86
87	### 3. Preserve raw output and build a bounded summary
88
89	```bash
90	MEMGRAPH=$(jq -er \
91	'select(.status == "captured") \| .memgraph \| select(type == "string" and length > 0)' \
92	/tmp/myapp-memory/run-01/capture.json)
93	test -s "$MEMGRAPH"
94	python3 scripts/summarize_memgraph.py \
95	"$MEMGRAPH" \
96	--artifact-dir /tmp/myapp-memory/run-01/analysis-raw \
97	--app-image 'MyApp\|MyFeatureKit' \
98	--trace-limit 3 --group-by-type --pretty \
99	> /tmp/myapp-memory/run-01/analysis.json
100	```
101
102	Read the exact graph path from the preserved capture report; do not guess a
103	timestamped filename. The helper creates a dedicated raw-artifact directory,
104	refuses to reuse it, runs `leaks --list`, and parses only a conservative subset
105	of its text. `--app-image` marks candidate rows; it does not prove ownership.
106	`--trace-limit` runs bounded
107	`leaks --traceTree=<address>` queries. Add `--reference-tree` when aggregate
108	root paths are more useful than individual leaked addresses. With
109	`--group-by-type`, that reference-tree query is grouped in the same invocation.
110	Exit statuses 0 and 1 from `

People who liked this also viewed

6 picks

Type

ios-memgraph-analysis

158installs45forks

View on GitHub

How to install ios-memgraph-analysis?

dpearson2699/swift-ios-skills/ios-memgraph-analysis

$npx -y skills add dpearson2699/swift-ios-skills --skill ios-memgraph-analysis

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# iOS Memgraph Analysis
2
3	Use memory graphs to prove why memory survives a defined lifetime boundary.
4	Separate unreachable leaks from reachable growth, preserve raw tool output, and
5	verify the same app-owned type and ownership path after a fix.
6
7	## Contents
8
9	- [Boundary](#boundary)
10	- [Evidence Model](#evidence-model)
11	- [Workflow](#workflow)
12	- [Ownership Decisions](#ownership-decisions)
13	- [Common Mistakes](#common-mistakes)
14	- [Review Checklist](#review-checklist)
15	- [References](#references)
16
17	## Boundary
18
19	This skill owns `.memgraph` capture and command-line ownership/growth analysis.
20	Use the Memory Graph Debugger or Instruments when their interactive graph and
21	allocation timeline are the primary task. Use source review for a suspected
22	closure capture only after runtime evidence identifies the lifetime or path.
23
24	## Evidence Model
25
26	Do not collapse these conditions:
27
28	- Unreachable leak: allocated memory no longer has a path from a live root.
29	An isolated strong cycle can be unreachable and still consume memory.
30	- Reachable but abandoned state: a live root still retains objects the user
31	flow no longer needs. `leaks` may correctly report zero.
32	- Expected cache or pool: memory survives intentionally and must be judged by
33	its bound, eviction behavior, and pressure response.
34	- Heap regression or fragmentation: footprint grows because more/larger
35	allocations persist or dirty pages are poorly utilized, without a leak.
36
37	Apple's leak scanner uses conservative pointer discovery and incomplete type
38	metadata. Counts can fluctuate, and a zero result does not prove the absence of
39	an ownership bug. Strong evidence identifies the expected lifetime, an
40	app-owned type or allocation, and a credible path or isolated reproduction.
41
42	## Workflow
43
44	### 1. Define the lifetime before capturing
45
46	Name the object that should disappear and the event that ends its useful life.
47	For example: `EditorViewModel` should deinitialize after dismissing the editor
48	and completing pending save work.
49
50	Record one deterministic sequence:
51
52	1. launch or restore a known state;
53	2. take an optional baseline graph;
54	3. perform the feature flow;
55	4. cross the expected release boundary;
56	5. wait for legitimate asynchronous cleanup;
57	6. take the post-flow graph.
58
59	Keep build, simulator/device, data, Malloc Stack Logging setting, and repetition
60	count stable. Malloc Stack Logging adds valuable allocation backtraces but also
61	overhead; compare only runs with the same setting.
62
63	### 2. Capture a graph without guessing the process
64
65	Xcode can export a graph from the Memory Graph Debugger. For a running Simulator
66	app, use the helper from this skill:
67
68	```bash
69	mkdir -p /tmp/myapp-memory
70	mkdir /tmp/myapp-memory/run-01
71	python3 scripts/capture_sim_memgraph.py \
72	--bundle-id com.example.MyApp \
73	--output-dir /tmp/myapp-memory/run-01 \
74	--pretty > /tmp/myapp-memory/run-01/capture.json
75	```
76
77	The per-run `mkdir` must fail if the capture directory already exists. Use a
78	new run name rather than mixing stale evidence with a retry.
79
80	Pass `--udid` when more than one Simulator is booted. The helper accepts only
81	one exact launchd label and PID; zero or multiple matches are errors. It runs the
82	host `leaks --outputGraph` command, retains stdout/stderr, and writes a manifest.
83	Do not replace this with `pgrep \| head -1` or a substring match.
84
85	Capturing suspends the process. Do not use capture latency as performance data.
86
87	### 3. Preserve raw output and build a bounded summary
88
89	```bash
90	MEMGRAPH=$(jq -er \
91	'select(.status == "captured") \| .memgraph \| select(type == "string" and length > 0)' \
92	/tmp/myapp-memory/run-01/capture.json)
93	test -s "$MEMGRAPH"
94	python3 scripts/summarize_memgraph.py \
95	"$MEMGRAPH" \
96	--artifact-dir /tmp/myapp-memory/run-01/analysis-raw \
97	--app-image 'MyApp\|MyFeatureKit' \
98	--trace-limit 3 --group-by-type --pretty \
99	> /tmp/myapp-memory/run-01/analysis.json
100	```
101
102	Read the exact graph path from the preserved capture report; do not guess a
103	timestamped filename. The helper creates a dedicated raw-artifact directory,
104	refuses to reuse it, runs `leaks --list`, and parses only a conservative subset
105	of its text. `--app-image` marks candidate rows; it does not prove ownership.
106	`--trace-limit` runs bounded
107	`leaks --traceTree=<address>` queries. Add `--reference-tree` when aggregate
108	root paths are more useful than individual leaked addresses. With
109	`--group-by-type`, that reference-tree query is grouped in the same invocation.
110	Exit statuses 0 and 1 from `

People who liked this also viewed

6 picks

Type