think-before-coding

Name: think-before-coding
Author: DevelopersGlobal

64stars10forks

How to install think-before-coding?

developersglobal/ai-agent-skills/think-before-coding

$npx -y skills add DevelopersGlobal/ai-agent-skills --skill think-before-coding

Installs into the current project.

Summary

Forces explicit reasoning before writing any code. Surfaces assumptions, manages confusion, and prevents hallucination by demanding clarity upfront.

Files · 1

View on GitHub

SKILL.md

1	## Overview
2
3	AI agents default to writing code immediately — even when the request is ambiguous, the context is incomplete, or multiple valid interpretations exist. This skill forces a deliberate thinking phase before any implementation begins.
4
5	The result: fewer rewrites, fewer wrong-direction implementations, and fewer "I assumed X" surprises.
6
7	## When to Use
8
9	Activate this skill before writing any code when:
10	- The request is ambiguous or underspecified
11	- Multiple valid implementations exist
12	- You are unsure about an existing system's constraints
13	- The task touches security, data integrity, or public APIs
14	- You feel the urge to "just start coding" to make progress
15
16	## Process
17
18	### Phase 1: Understand the Request
19
20	1. Read the full request — Do not skim. Re-read once.
21	2. Identify ambiguities — List every decision you would have to make silently:
22	- What counts as "success"?
23	- What are the inputs and expected outputs?
24	- What should NOT change?
25	- Are there existing patterns to follow?
26	3. Surface assumptions — Write them out: "I am assuming X because Y."
27	4. Check for contradictions — Does the request contradict itself or existing code?
28
29	Verify: Can you state the goal in one clear sentence? If not, ask before proceeding.
30
31	### Phase 2: Clarify Before Acting
32
33	5. If genuinely ambiguous, ask a focused clarifying question — not a list of 10 questions. Pick the one blocker.
34	6. If multiple valid approaches exist, present 2–3 options with tradeoffs, then ask which to proceed with.
35	7. If something seems wrong with the request, say so directly — don't silently work around it.
36
37	Verify: You have a clear, agreed-upon interpretation of the task.
38
39	### Phase 3: State Your Plan
40
41	8. Before writing code, state:
42	- What you will build
43	- What you will NOT touch
44	- What success looks like (your verification criteria)
45	9. For multi-step tasks, write a brief plan:
46	```
47	1. [Step] → verify: [check]
48	2. [Step] → verify: [check]
49	3. [Step] → verify: [check]
50	```
51
52	Verify: The plan is agreed upon, or you have received approval to proceed.
53
54	### Phase 4: Code
55
56	10. Now write the code — following the plan exactly.
57
58	## Common Rationalizations (and Rebuttals)
59
60	\| Excuse \| Rebuttal \|
61	\|--------\|----------\|
62	\| "I'll figure it out as I go" \| Unknown unknowns compound. 5 minutes of thinking saves 50 minutes of rewriting. \|
63	\| "The request is clear enough" \| If you can't state the goal in one sentence, it's not clear enough. \|
64	\| "Asking questions slows things down" \| Wrong direction is infinitely slower than a 30-second clarifying question. \|
65	\| "I'll handle edge cases later" \| Edge cases not considered upfront become bugs found in production. \|
66	\| "I can always refactor" \| You almost never will. Design now. \|
67
68	## Red Flags
69
70	- You're already writing code and you haven't stated what success looks like
71	- You made a silent assumption about what the user wants
72	- You chose between two approaches without mentioning the tradeoff
73	- You're solving a problem the user didn't ask about
74	- You're 100 lines in and realize you misunderstood the goal
75
76	## Verification
77
78	Before leaving this phase, confirm:
79	- [ ] The goal is stated in one clear sentence
80	- [ ] All major assumptions are explicitly named
81	- [ ] Ambiguities are resolved (or flagged for the user)
82	- [ ] A brief plan exists with verification criteria per step
83	- [ ] You know what you will NOT touch
84
85	## References
86
87	- [Andrej Karpathy on LLM coding pitfalls](https://x.com/karpathy/status/2015883857489522876)
88	- [goal-driven-execution skill](../goal-driven-execution/SKILL.md)
89	- [idea-to-spec skill](../idea-to-spec/SKILL.md)

People who liked this also viewed

6 picks

Type

hyperframes-cliUse the HyperFrames CLI development loop: init, add, catalog, capture, lint, check, snapshot, compare, grade-compare, preview, play, present, beats, keyframes, single or batch render, publish, cloud, SkillsJul 2026262k37k
request-refactor-planSkills for Real Engineers. Straight from my .agents directory.SkillsJul 2026165k181k
analyze-projectRigor Analyze / Rigor Audit read-only skill for deep learning research repositories. Use when the user wants to read and understand a repository, inspect model structure and training or inference entrSkillsMay 2026161k507
explore-codeRigor Improve implementation leaf skill for auditable candidate implementation in deep learning research repositories. Use when the researcher explicitly authorizes exploratory work on an isolated braSkillsMay 2026161k507
ubiquitous-languageSkills for Real Engineers. Straight from my .agents directory.SkillsJul 2026160k181k
scaffold-exercisesSkills for Real Engineers. Straight from my .agents directory.SkillsJul 2026149k181k

think-before-coding

64stars10forks

View on GitHub

How to install think-before-coding?

developersglobal/ai-agent-skills/think-before-coding

$npx -y skills add DevelopersGlobal/ai-agent-skills --skill think-before-coding

Installs into the current project.

Summary

Forces explicit reasoning before writing any code. Surfaces assumptions, manages confusion, and prevents hallucination by demanding clarity upfront.

Files · 1

View on GitHub

SKILL.md

1	## Overview
2
3	AI agents default to writing code immediately — even when the request is ambiguous, the context is incomplete, or multiple valid interpretations exist. This skill forces a deliberate thinking phase before any implementation begins.
4
5	The result: fewer rewrites, fewer wrong-direction implementations, and fewer "I assumed X" surprises.
6
7	## When to Use
8
9	Activate this skill before writing any code when:
10	- The request is ambiguous or underspecified
11	- Multiple valid implementations exist
12	- You are unsure about an existing system's constraints
13	- The task touches security, data integrity, or public APIs
14	- You feel the urge to "just start coding" to make progress
15
16	## Process
17
18	### Phase 1: Understand the Request
19
20	1. Read the full request — Do not skim. Re-read once.
21	2. Identify ambiguities — List every decision you would have to make silently:
22	- What counts as "success"?
23	- What are the inputs and expected outputs?
24	- What should NOT change?
25	- Are there existing patterns to follow?
26	3. Surface assumptions — Write them out: "I am assuming X because Y."
27	4. Check for contradictions — Does the request contradict itself or existing code?
28
29	Verify: Can you state the goal in one clear sentence? If not, ask before proceeding.
30
31	### Phase 2: Clarify Before Acting
32
33	5. If genuinely ambiguous, ask a focused clarifying question — not a list of 10 questions. Pick the one blocker.
34	6. If multiple valid approaches exist, present 2–3 options with tradeoffs, then ask which to proceed with.
35	7. If something seems wrong with the request, say so directly — don't silently work around it.
36
37	Verify: You have a clear, agreed-upon interpretation of the task.
38
39	### Phase 3: State Your Plan
40
41	8. Before writing code, state:
42	- What you will build
43	- What you will NOT touch
44	- What success looks like (your verification criteria)
45	9. For multi-step tasks, write a brief plan:
46	```
47	1. [Step] → verify: [check]
48	2. [Step] → verify: [check]
49	3. [Step] → verify: [check]
50	```
51
52	Verify: The plan is agreed upon, or you have received approval to proceed.
53
54	### Phase 4: Code
55
56	10. Now write the code — following the plan exactly.
57
58	## Common Rationalizations (and Rebuttals)
59
60	\| Excuse \| Rebuttal \|
61	\|--------\|----------\|
62	\| "I'll figure it out as I go" \| Unknown unknowns compound. 5 minutes of thinking saves 50 minutes of rewriting. \|
63	\| "The request is clear enough" \| If you can't state the goal in one sentence, it's not clear enough. \|
64	\| "Asking questions slows things down" \| Wrong direction is infinitely slower than a 30-second clarifying question. \|
65	\| "I'll handle edge cases later" \| Edge cases not considered upfront become bugs found in production. \|
66	\| "I can always refactor" \| You almost never will. Design now. \|
67
68	## Red Flags
69
70	- You're already writing code and you haven't stated what success looks like
71	- You made a silent assumption about what the user wants
72	- You chose between two approaches without mentioning the tradeoff
73	- You're solving a problem the user didn't ask about
74	- You're 100 lines in and realize you misunderstood the goal
75
76	## Verification
77
78	Before leaving this phase, confirm:
79	- [ ] The goal is stated in one clear sentence
80	- [ ] All major assumptions are explicitly named
81	- [ ] Ambiguities are resolved (or flagged for the user)
82	- [ ] A brief plan exists with verification criteria per step
83	- [ ] You know what you will NOT touch
84
85	## References
86
87	- [Andrej Karpathy on LLM coding pitfalls](https://x.com/karpathy/status/2015883857489522876)
88	- [goal-driven-execution skill](../goal-driven-execution/SKILL.md)
89	- [idea-to-spec skill](../idea-to-spec/SKILL.md)

People who liked this also viewed

6 picks

Type