think-in-code

Name: think-in-code
Author: fusengine

21stars3forks

How to install think-in-code?

fusengine/agents/think-in-code

$npx -y skills add fusengine/agents --skill think-in-code

Installs into the current project.

Summary

Use ONE Bash script instead of N sequential Read calls when analyzing multiple files, auditing codebase, finding all matches, scanning dependencies, counting lines, or listing files matching a pattern. Replaces wasteful multi-Read loops with compact shell pipelines.

Files · 1

View on GitHub

SKILL.md

1	# Think-in-Code Skill
2
3	## Principle
4
5	1 Bash script = N Read calls avoided.
6
7	When you'd read 10 files sequentially to extract a summary, you waste tokens loading full contents into context. Instead: 1 shell pipeline returns the compact aggregated result.
8
9	Heuristic: if your task is "for each file in set, compute/extract X, then aggregate" → write the script. Reserve Read for targeted inspection of a specific file you already know matters.
10
11	## 5 Runnable Patterns
12
13	### 1. File size audit (> 100 lines violations)
14
15	```bash
16	find . -type f $ -name ".ts" -o -name ".tsx" -o -name ".py" -o -name ".php" $ \
17	-not -path "/node_modules/" -not -path "/vendor/" \
18	\| xargs wc -l 2>/dev/null \
19	\| awk '$1 > 100 && $2 != "total" {print $1, $2}' \
20	\| sort -rn \| head -20
21	```
22
23	### 2. Multi-grep symbols (compact JSON)
24
25	```bash
26	rg --json -g '.ts' -g '.tsx' 'export (function\|class\|const) (\w+)' src/ \
27	\| jq -r 'select(.type=="match") \| "\(.data.path.text):\(.data.line_number) \(.data.lines.text)"' \
28	\| head -50
29	```
30
31	### 3. Dependencies with versions
32
33	```bash
34	# Node
35	jq -r '.dependencies, .devDependencies \| to_entries[] \| "\(.key)@\(.value)"' package.json 2>/dev/null
36
37	# PHP
38	jq -r '.require, ."require-dev" \| to_entries[] \| "\(.key)@\(.value)"' composer.json 2>/dev/null
39	```
40
41	### 4. Error log scan
42
43	```bash
44	grep -rEn 'ERROR\|FATAL\|Exception\|panic:\|stack trace' \
45	--include="*.log" logs/ 2>/dev/null \
46	\| tail -30
47	```
48
49	### 5. Lines of code by extension
50
51	```bash
52	find . -type f $ -name ".ts" -o -name ".tsx" -o -name ".js" -o -name ".py" -o -name "*.php" $ \
53	-not -path "/node_modules/" -not -path "/vendor/" -not -path "/.git/" \
54	-exec wc -l {} + \
55	\| awk '{ext=$2; sub(/.*\./,"",ext); sum[ext]+=$1} END {for (e in sum) print sum[e], e}' \
56	\| sort -rn
57	```
58
59	## Anti-Pattern
60
61	Do NOT Read 10 files sequentially to count their lines, list their exports, or check their size. That is ~6KB context per file = ~60KB wasted for a result that fits in 1KB.
62
63	```
64	WRONG: Read(f1.ts) → Read(f2.ts) → ... → Read(f10.ts) → manual tally
65	RIGHT: Bash(find ... \| xargs wc -l \| awk ...) → 1 compact table
66	```
67
68	## Before / After
69
70	Task: "Find files > 100 lines in `src/`."
71
72	- Before: 10 × Read full file → ~60KB tokens consumed, then mental wc.
73	- After: 1 × `find src -name '*.ts' \| xargs wc -l \| awk '$1>100'` → ~1KB result.
74
75	~60× reduction. Use the script.

People who liked this also viewed

6 picks

Type

think-in-code

21stars3forks

View on GitHub

How to install think-in-code?

fusengine/agents/think-in-code

$npx -y skills add fusengine/agents --skill think-in-code

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Think-in-Code Skill
2
3	## Principle
4
5	1 Bash script = N Read calls avoided.
6
7	When you'd read 10 files sequentially to extract a summary, you waste tokens loading full contents into context. Instead: 1 shell pipeline returns the compact aggregated result.
8
9	Heuristic: if your task is "for each file in set, compute/extract X, then aggregate" → write the script. Reserve Read for targeted inspection of a specific file you already know matters.
10
11	## 5 Runnable Patterns
12
13	### 1. File size audit (> 100 lines violations)
14
15	```bash
16	find . -type f $ -name ".ts" -o -name ".tsx" -o -name ".py" -o -name ".php" $ \
17	-not -path "/node_modules/" -not -path "/vendor/" \
18	\| xargs wc -l 2>/dev/null \
19	\| awk '$1 > 100 && $2 != "total" {print $1, $2}' \
20	\| sort -rn \| head -20
21	```
22
23	### 2. Multi-grep symbols (compact JSON)
24
25	```bash
26	rg --json -g '.ts' -g '.tsx' 'export (function\|class\|const) (\w+)' src/ \
27	\| jq -r 'select(.type=="match") \| "\(.data.path.text):\(.data.line_number) \(.data.lines.text)"' \
28	\| head -50
29	```
30
31	### 3. Dependencies with versions
32
33	```bash
34	# Node
35	jq -r '.dependencies, .devDependencies \| to_entries[] \| "\(.key)@\(.value)"' package.json 2>/dev/null
36
37	# PHP
38	jq -r '.require, ."require-dev" \| to_entries[] \| "\(.key)@\(.value)"' composer.json 2>/dev/null
39	```
40
41	### 4. Error log scan
42
43	```bash
44	grep -rEn 'ERROR\|FATAL\|Exception\|panic:\|stack trace' \
45	--include="*.log" logs/ 2>/dev/null \
46	\| tail -30
47	```
48
49	### 5. Lines of code by extension
50
51	```bash
52	find . -type f $ -name ".ts" -o -name ".tsx" -o -name ".js" -o -name ".py" -o -name "*.php" $ \
53	-not -path "/node_modules/" -not -path "/vendor/" -not -path "/.git/" \
54	-exec wc -l {} + \
55	\| awk '{ext=$2; sub(/.*\./,"",ext); sum[ext]+=$1} END {for (e in sum) print sum[e], e}' \
56	\| sort -rn
57	```
58
59	## Anti-Pattern
60
61	Do NOT Read 10 files sequentially to count their lines, list their exports, or check their size. That is ~6KB context per file = ~60KB wasted for a result that fits in 1KB.
62
63	```
64	WRONG: Read(f1.ts) → Read(f2.ts) → ... → Read(f10.ts) → manual tally
65	RIGHT: Bash(find ... \| xargs wc -l \| awk ...) → 1 compact table
66	```
67
68	## Before / After
69
70	Task: "Find files > 100 lines in `src/`."
71
72	- Before: 10 × Read full file → ~60KB tokens consumed, then mental wc.
73	- After: 1 × `find src -name '*.ts' \| xargs wc -l \| awk '$1>100'` → ~1KB result.
74
75	~60× reduction. Use the script.

People who liked this also viewed

6 picks

Type