prompt-master

Name: prompt-master
Author: pavel-molyanov

241stars75forks

How to install prompt-master?

pavel-molyanov/molyanov-ai-dev/prompt-master

$npx -y skills add pavel-molyanov/molyanov-ai-dev --skill prompt-master

Installs into the current project.

Summary

Guide for writing effective prompts for LLMs. Use when: "напиши промпт", "улучши промпт", "prompt engineering", "проверь промпт

Files · 1

View on GitHub

SKILL.md

1	<!-- Generated by sync-to-codex v1. Do not edit directly. -->
2
3	# Prompt Engineering for Reasoning Models
4
5	Based on Anthropic and OpenAI guidelines (2025-2026). Every principle here has a motivation — when you understand WHY something works, you follow it more reliably.
6
7	## Core Principles
8
9	### The model is already smart
10
11	Add only what the model lacks: domain context, constraints, success criteria. Every sentence should justify its token cost. The context window is a shared resource — prompts compete for attention with conversation history, tool outputs, and the model's own reasoning.
12
13	### Clarity over cleverness
14
15	Most prompt failures stem from ambiguity, not model limitations. Test: show the prompt to a colleague with no context. If they're confused about what to do, the model will be too.
16
17	### Motivation over emphasis
18
19	Explain WHY a rule matters. One motivated sentence outperforms ten capitalized words.
20
21	When every instruction screams for attention (ALL CAPS, "CRITICAL", "NEVER", "ALWAYS", "MUST"), nothing stands out. Emphasis words signal a poorly written instruction — rewrite it instead of raising the volume. Over-thorough language ("Be THOROUGH", "Make sure you have the FULL picture") also hurts — it inflates token cost without adding signal.
22
23	```
24	Before:
25	CRITICAL: You MUST ALWAYS validate input. NEVER skip validation.
26	IMPORTANT: ALWAYS check for edge cases. This is MANDATORY.
27
28	After:
29	Validate all input before processing.
30	Reason: unvalidated input causes pipeline crashes in production.
31	```
32
33	### Positive framing
34
35	Default to stating what you want, not what to avoid. Models follow positive instructions more reliably. Long prohibition lists get ignored — they add tokens without adding clarity.
36
37	```
38	Rewrite when positive form is sufficient:
39	Before: Don't use bullet points. Never include code examples.
40	After: Write in prose paragraphs, 2-3 sentences each.
41
42	Keep negatives for hard boundaries where positive rewrite loses the prohibition:
43	"Return data as JSON. Do not include markdown fences around it."
44	— positive alone ("Return raw JSON") may not prevent the common mistake.
45
46	Test: does the positive rewrite fully convey the prohibition?
47	Yes → rewrite positively.
48	No → keep negative + add reason why it matters.
49	```
50
51	### Examples over rules
52
53	1-3 canonical examples transfer knowledge more efficiently than paragraphs of description. Show the desired output — let the model generalize from the pattern.
54
55	### Compress
56
57	Remove filler ("could you please", "I would like you to", "make sure to"). Shorter prompts often perform equally well or better — less noise means stronger signal per token.
58
59	### Degrees of freedom
60
61	Match specificity to the task's fragility. Over-specifying creative tasks stifles the model's reasoning. Under-specifying fragile tasks leads to format errors and broken parsing.
62
63	Fragile tasks (parsing specific formats, following exact protocols): prescribe steps.
64	Creative tasks (writing, analysis, design): give constraints and let the model find its path.
65
66	### High-level guidance
67
68	Provide constraints and success criteria rather than step-by-step micro-instructions. The model's own reasoning often exceeds prescriptive procedures. Describe WHAT success looks like — the model figures out HOW.
69
70	## Techniques That Work
71
72	### Context over decoration
73
74	Provide concrete context (audience, use case, constraints) instead of decorative phrasing. The model gains nothing from flattery.
75
76	```
77	Before:
78	You are an incredibly brilliant and talented expert programmer
79	who writes the most amazing code in the world. Please write
80	a function that validates emails.
81
82	After:
83	Write an email validation function.
84	Context: TypeScript, used in a signup form, must handle
85	international domains. Return {valid: boolean, reason: string}.
86	```
87
88	### Structure with XML tags
89
90	Separate instructions, data, and examples with XML tags. This prevents the model from confusing context with instructions. Consistent tag names make handoffs between chained prompts clean.
91
92	```
93	Before:
94	Here's some customer feedback. Also, the format should be
95	JSON. And please analyze sentiment. The feedback is:
96	"Great product but shipping was slow."
97
98	After:
99	<instructions>
100	Analyze sentiment. Output JSON: {text, sentiment, confidence}.
101	</instructions>
102	<data>
103	Great product but shipping was slow.
104	</data>
105	```
106
107	### Examples over explanations
108
109	Instead of describing the desired output format in paragraphs, show 1-3 examples. The model generalizes from patterns faster than from rules.
110
111	```
112	Before:
113	When summarizing articles, create a summary that includes
114	the main topic as a short phrase, then 2-3 key points as
115	bullet items, then a one-sentence practical takeaway.
116
117	After:
118	Summarize articles in this format:
119
120	<example>
121	Topic: Remote work productivity
122	- Asy

People who liked this also viewed

6 picks

Type

lark-drive飞书云空间（云盘/云存储）：管理 Drive 文件和文件夹，包含上传/下载、创建文件夹、复制/移动/删除、查看元数据、评论/权限/订阅、标题、版本、飞书文档密级标签（secure labels）和本地文件导入。用户需要整理云盘目录、处理云空间资源 URL/token、判断链接类型/真实 token/标题，或导入 Word/Markdown/Excel/CSV/PPTX/.base 为 docx/sSkillsJul 2026382k16k
lark-sharedUse for lark-cli setup/auth tasks: auth login/status/logout, user vs bot identity, business-domain permissions (--domain, including all/docs/drive), missing scopes, revoking authorization, or handlingSkillsJul 2026382k16k
lark-whiteboard飞书画板：查询和编辑飞书云文档中的画板。支持导出画板为预览图片、导出原始节点结构、使用多种格式更新画板内容。当用户需要查看画板内容、导出画板图片、编辑画板时使用此 skill。不负责：飞书云文档内容编辑（lark-doc）、文档内嵌电子表格/Base（lark-sheets / lark-base）。SkillsJul 2026380k16k
lark-task飞书任务：管理任务、清单和任务智能体。创建待办任务、查看和更新任务状态、拆分子任务、组织任务清单、分配协作成员、上传任务附件、注册或注销任务智能体、更新任务智能体的主页数据、写入智能体任务记录。当用户需要创建待办事项、查看任务列表、跟踪任务进度、管理项目清单或给他人分配任务、为任务上传附件文件、注册注销任务智能体、更新智能体主页数据、写入任务记录时使用。SkillsJul 2026380k16k
lark-attendance飞书考勤打卡：查询自己的考勤打卡记录SkillsJul 2026330k16k
lark-apps妙搭（Spark/Miaoda）应用开发与托管：应用创建、本地全栈开发、云端生成迭代、创意设计（UI mockup / 可交互原型 / 线框图 / 落地页 / 仪表盘 / 幻灯片 deck / 视觉探索）、AI相关能力和飞书平台能力或者其他外部能力集成、日志/Trace/监控指标/PV/UV 查询、环境变量管理、应用角色与成员管理、自动化触发器（定时/记录变更/Webhook/飞书审批）。当用户要SkillsJul 2026227k16k

prompt-master

241stars75forks

View on GitHub

How to install prompt-master?

pavel-molyanov/molyanov-ai-dev/prompt-master

$npx -y skills add pavel-molyanov/molyanov-ai-dev --skill prompt-master

Installs into the current project.

Summary

Guide for writing effective prompts for LLMs. Use when: "напиши промпт", "улучши промпт", "prompt engineering", "проверь промпт

Files · 1

View on GitHub

SKILL.md

1	<!-- Generated by sync-to-codex v1. Do not edit directly. -->
2
3	# Prompt Engineering for Reasoning Models
4
5	Based on Anthropic and OpenAI guidelines (2025-2026). Every principle here has a motivation — when you understand WHY something works, you follow it more reliably.
6
7	## Core Principles
8
9	### The model is already smart
10
11	Add only what the model lacks: domain context, constraints, success criteria. Every sentence should justify its token cost. The context window is a shared resource — prompts compete for attention with conversation history, tool outputs, and the model's own reasoning.
12
13	### Clarity over cleverness
14
15	Most prompt failures stem from ambiguity, not model limitations. Test: show the prompt to a colleague with no context. If they're confused about what to do, the model will be too.
16
17	### Motivation over emphasis
18
19	Explain WHY a rule matters. One motivated sentence outperforms ten capitalized words.
20
21	When every instruction screams for attention (ALL CAPS, "CRITICAL", "NEVER", "ALWAYS", "MUST"), nothing stands out. Emphasis words signal a poorly written instruction — rewrite it instead of raising the volume. Over-thorough language ("Be THOROUGH", "Make sure you have the FULL picture") also hurts — it inflates token cost without adding signal.
22
23	```
24	Before:
25	CRITICAL: You MUST ALWAYS validate input. NEVER skip validation.
26	IMPORTANT: ALWAYS check for edge cases. This is MANDATORY.
27
28	After:
29	Validate all input before processing.
30	Reason: unvalidated input causes pipeline crashes in production.
31	```
32
33	### Positive framing
34
35	Default to stating what you want, not what to avoid. Models follow positive instructions more reliably. Long prohibition lists get ignored — they add tokens without adding clarity.
36
37	```
38	Rewrite when positive form is sufficient:
39	Before: Don't use bullet points. Never include code examples.
40	After: Write in prose paragraphs, 2-3 sentences each.
41
42	Keep negatives for hard boundaries where positive rewrite loses the prohibition:
43	"Return data as JSON. Do not include markdown fences around it."
44	— positive alone ("Return raw JSON") may not prevent the common mistake.
45
46	Test: does the positive rewrite fully convey the prohibition?
47	Yes → rewrite positively.
48	No → keep negative + add reason why it matters.
49	```
50
51	### Examples over rules
52
53	1-3 canonical examples transfer knowledge more efficiently than paragraphs of description. Show the desired output — let the model generalize from the pattern.
54
55	### Compress
56
57	Remove filler ("could you please", "I would like you to", "make sure to"). Shorter prompts often perform equally well or better — less noise means stronger signal per token.
58
59	### Degrees of freedom
60
61	Match specificity to the task's fragility. Over-specifying creative tasks stifles the model's reasoning. Under-specifying fragile tasks leads to format errors and broken parsing.
62
63	Fragile tasks (parsing specific formats, following exact protocols): prescribe steps.
64	Creative tasks (writing, analysis, design): give constraints and let the model find its path.
65
66	### High-level guidance
67
68	Provide constraints and success criteria rather than step-by-step micro-instructions. The model's own reasoning often exceeds prescriptive procedures. Describe WHAT success looks like — the model figures out HOW.
69
70	## Techniques That Work
71
72	### Context over decoration
73
74	Provide concrete context (audience, use case, constraints) instead of decorative phrasing. The model gains nothing from flattery.
75
76	```
77	Before:
78	You are an incredibly brilliant and talented expert programmer
79	who writes the most amazing code in the world. Please write
80	a function that validates emails.
81
82	After:
83	Write an email validation function.
84	Context: TypeScript, used in a signup form, must handle
85	international domains. Return {valid: boolean, reason: string}.
86	```
87
88	### Structure with XML tags
89
90	Separate instructions, data, and examples with XML tags. This prevents the model from confusing context with instructions. Consistent tag names make handoffs between chained prompts clean.
91
92	```
93	Before:
94	Here's some customer feedback. Also, the format should be
95	JSON. And please analyze sentiment. The feedback is:
96	"Great product but shipping was slow."
97
98	After:
99	<instructions>
100	Analyze sentiment. Output JSON: {text, sentiment, confidence}.
101	</instructions>
102	<data>
103	Great product but shipping was slow.
104	</data>
105	```
106
107	### Examples over explanations
108
109	Instead of describing the desired output format in paragraphs, show 1-3 examples. The model generalizes from patterns faster than from rules.
110
111	```
112	Before:
113	When summarizing articles, create a summary that includes
114	the main topic as a short phrase, then 2-3 key points as
115	bullet items, then a one-sentence practical takeaway.
116
117	After:
118	Summarize articles in this format:
119
120	<example>
121	Topic: Remote work productivity
122	- Asy

People who liked this also viewed

6 picks

Type