finding-duplicate-functions

Name: finding-duplicate-functions
Author: obra

535installs30forks

How to install finding-duplicate-functions?

obra/superpowers-lab/finding-duplicate-functions

$npx -y skills add obra/superpowers-lab --skill finding-duplicate-functions

Installs into the current project.

Summary

Use when auditing a codebase for semantic duplication - functions that do the same thing but have different names or implementations. Especially useful for LLM-generated codebases where new functions are often created rather than reusing existing ones.

Files · 1

View on GitHub

SKILL.md

1	# Finding Duplicate-Intent Functions
2
3	## Overview
4
5	LLM-generated codebases accumulate semantic duplicates: functions that serve the same purpose but were implemented independently. Classical copy-paste detectors (jscpd) find syntactic duplicates but miss "same intent, different implementation."
6
7	This skill uses a two-phase approach: classical extraction followed by LLM-powered intent clustering.
8
9	## When to Use
10
11	- Codebase has grown organically with multiple contributors (human or LLM)
12	- You suspect utility functions have been reimplemented multiple times
13	- Before major refactoring to identify consolidation opportunities
14	- After jscpd has been run and syntactic duplicates are already handled
15
16	## Quick Reference
17
18	\| Phase \| Tool \| Model \| Output \|
19	\|-------\|------\|-------\|--------\|
20	\| 1. Extract \| `scripts/extract-functions.sh` \| - \| `catalog.json` \|
21	\| 2. Categorize \| `scripts/categorize-prompt.md` \| haiku \| `categorized.json` \|
22	\| 3. Split \| `scripts/prepare-category-analysis.sh` \| - \| `categories/*.json` \|
23	\| 4. Detect \| `scripts/find-duplicates-prompt.md` \| opus \| `duplicates/*.json` \|
24	\| 5. Report \| `scripts/generate-report.sh` \| - \| `report.md` \|
25
26	## Process
27
28	```dot
29	digraph duplicate_detection {
30	rankdir=TB;
31	node [shape=box];
32
33	extract [label="1. Extract function catalog\n./scripts/extract-functions.sh"];
34	categorize [label="2. Categorize by domain\n(haiku subagent)"];
35	split [label="3. Split into categories\n./scripts/prepare-category-analysis.sh"];
36	detect [label="4. Find duplicates per category\n(opus subagent per category)"];
37	report [label="5. Generate report\n./scripts/generate-report.sh"];
38	review [label="6. Human review & consolidate"];
39
40	extract -> categorize -> split -> detect -> report -> review;
41	}
42	```
43
44	### Phase 1: Extract Function Catalog
45
46	```bash
47	./scripts/extract-functions.sh src/ -o catalog.json
48	```
49
50	Options:
51	- `-o FILE`: Output file (default: stdout)
52	- `-c N`: Lines of context to capture (default: 15)
53	- `-t GLOB`: File types (default: `.ts,.tsx,.js,.jsx`)
54	- `--include-tests`: Include test files (excluded by default)
55
56	Test files (`.test.`, `.spec.`, `__tests__/**`) are excluded by default since test utilities are less likely to be consolidation candidates.
57
58	### Phase 2: Categorize by Domain
59
60	Dispatch a haiku subagent using the prompt in `scripts/categorize-prompt.md`.
61
62	Insert the contents of `catalog.json` where indicated in the prompt template. Save output as `categorized.json`.
63
64	### Phase 3: Split into Categories
65
66	```bash
67	./scripts/prepare-category-analysis.sh categorized.json ./categories
68	```
69
70	Creates one JSON file per category. Only categories with 3+ functions are worth analyzing.
71
72	### Phase 4: Find Duplicates (Per Category)
73
74	For each category file in `./categories/`, dispatch an opus subagent using the prompt in `scripts/find-duplicates-prompt.md`.
75
76	Save each output as `./duplicates/{category}.json`.
77
78	### Phase 5: Generate Report
79
80	```bash
81	./scripts/generate-report.sh ./duplicates ./duplicates-report.md
82	```
83
84	Produces a prioritized markdown report grouped by confidence level.
85
86	### Phase 6: Human Review
87
88	Review the report. For HIGH confidence duplicates:
89	1. Verify the recommended survivor has tests
90	2. Update callers to use the survivor
91	3. Delete the duplicates
92	4. Run tests
93
94	## High-Risk Duplicate Zones
95
96	Focus extraction on these areas first - they accumulate duplicates fastest:
97
98	\| Zone \| Common Duplicates \|
99	\|------\|-------------------\|
100	\| `utils/`, `helpers/`, `lib/` \| General utilities reimplemented \|
101	\| Validation code \| Same checks written multiple ways \|
102	\| Error formatting \| Error-to-string conversions \|
103	\| Path manipulation \| Joining, resolving, normalizing paths \|
104	\| String formatting \| Case conversion, truncation, escaping \|
105	\| Date formatting \| Same formats implemented repeatedly \|
106	\| API response shaping \| Similar transformations for different endpoints \|
107
108	## Common Mistakes
109
110	Extracting too much: Focus on exported functions and public methods. Internal helpers are less likely to be duplicated across files.
111
112	Skipping the categorization step: Going straight to duplicate detection on the full catalog produces noise. Categories focus the comparison.
113
114	Using haiku for duplicate detection: Haiku is cost-effective for categorization but misses subtle semantic duplicates. Use Opus for the actual duplicate analysis.
115
116	Consolidating without tests: Before deleting duplicates, ensure the survivor has tests covering all use cases of the deleted functions.

People who liked this also viewed

6 picks

Type

finding-duplicate-functions

535installs30forks

View on GitHub

How to install finding-duplicate-functions?

obra/superpowers-lab/finding-duplicate-functions

$npx -y skills add obra/superpowers-lab --skill finding-duplicate-functions

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Finding Duplicate-Intent Functions
2
3	## Overview
4
5	LLM-generated codebases accumulate semantic duplicates: functions that serve the same purpose but were implemented independently. Classical copy-paste detectors (jscpd) find syntactic duplicates but miss "same intent, different implementation."
6
7	This skill uses a two-phase approach: classical extraction followed by LLM-powered intent clustering.
8
9	## When to Use
10
11	- Codebase has grown organically with multiple contributors (human or LLM)
12	- You suspect utility functions have been reimplemented multiple times
13	- Before major refactoring to identify consolidation opportunities
14	- After jscpd has been run and syntactic duplicates are already handled
15
16	## Quick Reference
17
18	\| Phase \| Tool \| Model \| Output \|
19	\|-------\|------\|-------\|--------\|
20	\| 1. Extract \| `scripts/extract-functions.sh` \| - \| `catalog.json` \|
21	\| 2. Categorize \| `scripts/categorize-prompt.md` \| haiku \| `categorized.json` \|
22	\| 3. Split \| `scripts/prepare-category-analysis.sh` \| - \| `categories/*.json` \|
23	\| 4. Detect \| `scripts/find-duplicates-prompt.md` \| opus \| `duplicates/*.json` \|
24	\| 5. Report \| `scripts/generate-report.sh` \| - \| `report.md` \|
25
26	## Process
27
28	```dot
29	digraph duplicate_detection {
30	rankdir=TB;
31	node [shape=box];
32
33	extract [label="1. Extract function catalog\n./scripts/extract-functions.sh"];
34	categorize [label="2. Categorize by domain\n(haiku subagent)"];
35	split [label="3. Split into categories\n./scripts/prepare-category-analysis.sh"];
36	detect [label="4. Find duplicates per category\n(opus subagent per category)"];
37	report [label="5. Generate report\n./scripts/generate-report.sh"];
38	review [label="6. Human review & consolidate"];
39
40	extract -> categorize -> split -> detect -> report -> review;
41	}
42	```
43
44	### Phase 1: Extract Function Catalog
45
46	```bash
47	./scripts/extract-functions.sh src/ -o catalog.json
48	```
49
50	Options:
51	- `-o FILE`: Output file (default: stdout)
52	- `-c N`: Lines of context to capture (default: 15)
53	- `-t GLOB`: File types (default: `.ts,.tsx,.js,.jsx`)
54	- `--include-tests`: Include test files (excluded by default)
55
56	Test files (`.test.`, `.spec.`, `__tests__/**`) are excluded by default since test utilities are less likely to be consolidation candidates.
57
58	### Phase 2: Categorize by Domain
59
60	Dispatch a haiku subagent using the prompt in `scripts/categorize-prompt.md`.
61
62	Insert the contents of `catalog.json` where indicated in the prompt template. Save output as `categorized.json`.
63
64	### Phase 3: Split into Categories
65
66	```bash
67	./scripts/prepare-category-analysis.sh categorized.json ./categories
68	```
69
70	Creates one JSON file per category. Only categories with 3+ functions are worth analyzing.
71
72	### Phase 4: Find Duplicates (Per Category)
73
74	For each category file in `./categories/`, dispatch an opus subagent using the prompt in `scripts/find-duplicates-prompt.md`.
75
76	Save each output as `./duplicates/{category}.json`.
77
78	### Phase 5: Generate Report
79
80	```bash
81	./scripts/generate-report.sh ./duplicates ./duplicates-report.md
82	```
83
84	Produces a prioritized markdown report grouped by confidence level.
85
86	### Phase 6: Human Review
87
88	Review the report. For HIGH confidence duplicates:
89	1. Verify the recommended survivor has tests
90	2. Update callers to use the survivor
91	3. Delete the duplicates
92	4. Run tests
93
94	## High-Risk Duplicate Zones
95
96	Focus extraction on these areas first - they accumulate duplicates fastest:
97
98	\| Zone \| Common Duplicates \|
99	\|------\|-------------------\|
100	\| `utils/`, `helpers/`, `lib/` \| General utilities reimplemented \|
101	\| Validation code \| Same checks written multiple ways \|
102	\| Error formatting \| Error-to-string conversions \|
103	\| Path manipulation \| Joining, resolving, normalizing paths \|
104	\| String formatting \| Case conversion, truncation, escaping \|
105	\| Date formatting \| Same formats implemented repeatedly \|
106	\| API response shaping \| Similar transformations for different endpoints \|
107
108	## Common Mistakes
109
110	Extracting too much: Focus on exported functions and public methods. Internal helpers are less likely to be duplicated across files.
111
112	Skipping the categorization step: Going straight to duplicate detection on the full catalog produces noise. Categories focus the comparison.
113
114	Using haiku for duplicate detection: Haiku is cost-effective for categorization but misses subtle semantic duplicates. Use Opus for the actual duplicate analysis.
115
116	Consolidating without tests: Before deleting duplicates, ensure the survivor has tests covering all use cases of the deleted functions.

People who liked this also viewed

6 picks

Type