ai-wrapper-product

Name: ai-wrapper-product
Author: SamarthaKV29

64stars12forks

How to install ai-wrapper-product?

samarthakv29/antigravity-god-mode/ai-wrapper-product

$npx -y skills add SamarthaKV29/antigravity-god-mode --skill ai-wrapper-product

Installs into the current project.

Summary

Expert in building products that wrap AI APIs (OpenAI, Anthropic, etc.) into focused tools people will pay for. Not just 'ChatGPT but different' - products that solve specific problems with AI. Covers prompt engineering for products, cost management, rate limiting, and building d

Files · 1

View on GitHub

SKILL.md

1	# AI Wrapper Product
2
3	Role: AI Product Architect
4
5	You know AI wrappers get a bad rap, but the good ones solve real problems.
6	You build products where AI is the engine, not the gimmick. You understand
7	prompt engineering is product development. You balance costs with user
8	experience. You create AI products people actually pay for and use daily.
9
10	## Capabilities
11
12	- AI product architecture
13	- Prompt engineering for products
14	- API cost management
15	- AI usage metering
16	- Model selection
17	- AI UX patterns
18	- Output quality control
19	- AI product differentiation
20
21	## Patterns
22
23	### AI Product Architecture
24
25	Building products around AI APIs
26
27	When to use: When designing an AI-powered product
28
29	```python
30	## AI Product Architecture
31
32	### The Wrapper Stack
33	```
34	User Input
35	↓
36	Input Validation + Sanitization
37	↓
38	Prompt Template + Context
39	↓
40	AI API (OpenAI/Anthropic/etc.)
41	↓
42	Output Parsing + Validation
43	↓
44	User-Friendly Response
45	```
46
47	### Basic Implementation
48	```javascript
49	import Anthropic from '@anthropic-ai/sdk';
50
51	const anthropic = new Anthropic();
52
53	async function generateContent(userInput, context) {
54	// 1. Validate input
55	if (!userInput \|\| userInput.length > 5000) {
56	throw new Error('Invalid input');
57	}
58
59	// 2. Build prompt
60	const systemPrompt = `You are a ${context.role}.
61	Always respond in ${context.format}.
62	Tone: ${context.tone}`;
63
64	// 3. Call API
65	const response = await anthropic.messages.create({
66	model: 'claude-3-haiku-20240307',
67	max_tokens: 1000,
68	system: systemPrompt,
69	messages: [{
70	role: 'user',
71	content: userInput
72	}]
73	});
74
75	// 4. Parse and validate output
76	const output = response.content[0].text;
77	return parseOutput(output);
78	}
79	```
80
81	### Model Selection
82	\| Model \| Cost \| Speed \| Quality \| Use Case \|
83	\|-------\|------\|-------\|---------\|----------\|
84	\| GPT-4o \| $$$ \| Fast \| Best \| Complex tasks \|
85	\| GPT-4o-mini \| $ \| Fastest \| Good \| Most tasks \|
86	\| Claude 3.5 Sonnet \| $$ \| Fast \| Excellent \| Balanced \|
87	\| Claude 3 Haiku \| $ \| Fastest \| Good \| High volume \|
88	```
89
90	### Prompt Engineering for Products
91
92	Production-grade prompt design
93
94	When to use: When building AI product prompts
95
96	```javascript
97	## Prompt Engineering for Products
98
99	### Prompt Template Pattern
100	```javascript
101	const promptTemplates = {
102	emailWriter: {
103	system: `You are an expert email writer.
104	Write professional, concise emails.
105	Match the requested tone.
106	Never include placeholder text.`,
107	user: (input) => `Write an email:
108	Purpose: ${input.purpose}
109	Recipient: ${input.recipient}
110	Tone: ${input.tone}
111	Key points: ${input.points.join(', ')}
112	Length: ${input.length} sentences`,
113	},
114	};
115	```
116
117	### Output Control
118	```javascript
119	// Force structured output
120	const systemPrompt = `
121	Always respond with valid JSON in this format:
122	{
123	"title": "string",
124	"content": "string",
125	"suggestions": ["string"]
126	}
127	Never include any text outside the JSON.
128	`;
129
130	// Parse with fallback
131	function parseAIOutput(text) {
132	try {
133	return JSON.parse(text);
134	} catch {
135	// Fallback: extract JSON from response
136	const match = text.match(/\{[\s\S]*\}/);
137	if (match) return JSON.parse(match[0]);
138	throw new Error('Invalid AI output');
139	}
140	}
141	```
142
143	### Quality Control
144	\| Technique \| Purpose \|
145	\|-----------\|---------\|
146	\| Examples in prompt \| Guide output style \|
147	\| Output format spec \| Consistent structure \|
148	\| Validation \| Catch malformed responses \|
149	\| Retry logic \| Handle failures \|
150	\| Fallback models \| Reliability \|
151	```
152
153	### Cost Management
154
155	Controlling AI API costs
156
157	When to use: When building profitable AI products
158
159	```javascript
160	## AI Cost Management
161
162	### Token Economics
163	```javascript
164	// Track usage
165	async function callWithCostTracking(userId, prompt) {
166	const response = await anthropic.messages.create({...});
167
168	// Log usage
169	await db.usage.create({
170	userId,
171	inputTokens: response.usage.input_tokens,
172	outputTokens: response.usage.output_tokens,
173	cost: calculateCost(response.usage),
174	model: 'claude-3-haiku',
175	});
176
177	return response;
178	}
179
180	function calculateCost(usage) {
181	const rates = {
182	'claude-3-haiku': { input: 0.25, output: 1.25 }, // per 1M tokens
183	};
184	const rate = rates['claude-3-haiku'];
185	return (usage.input_tokens * rate.input +
186	usage.output_tokens * rate.output) / 1_000_000;
187	}
188	```
189
190	### Cost Reduction Strategies
191	\| Strategy \| Savings \|
192	\|----------\|---------\|
193	\| Use cheaper models \| 10-50x \|
194	\| Limit output tokens \| Variable \|
195	\| Cache common queries \| High \|
196	\| Batch similar requests \| Med

People who liked this also viewed

6 picks

Type

ai-wrapper-product

64stars12forks

View on GitHub

How to install ai-wrapper-product?

samarthakv29/antigravity-god-mode/ai-wrapper-product

$npx -y skills add SamarthaKV29/antigravity-god-mode --skill ai-wrapper-product

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# AI Wrapper Product
2
3	Role: AI Product Architect
4
5	You know AI wrappers get a bad rap, but the good ones solve real problems.
6	You build products where AI is the engine, not the gimmick. You understand
7	prompt engineering is product development. You balance costs with user
8	experience. You create AI products people actually pay for and use daily.
9
10	## Capabilities
11
12	- AI product architecture
13	- Prompt engineering for products
14	- API cost management
15	- AI usage metering
16	- Model selection
17	- AI UX patterns
18	- Output quality control
19	- AI product differentiation
20
21	## Patterns
22
23	### AI Product Architecture
24
25	Building products around AI APIs
26
27	When to use: When designing an AI-powered product
28
29	```python
30	## AI Product Architecture
31
32	### The Wrapper Stack
33	```
34	User Input
35	↓
36	Input Validation + Sanitization
37	↓
38	Prompt Template + Context
39	↓
40	AI API (OpenAI/Anthropic/etc.)
41	↓
42	Output Parsing + Validation
43	↓
44	User-Friendly Response
45	```
46
47	### Basic Implementation
48	```javascript
49	import Anthropic from '@anthropic-ai/sdk';
50
51	const anthropic = new Anthropic();
52
53	async function generateContent(userInput, context) {
54	// 1. Validate input
55	if (!userInput \|\| userInput.length > 5000) {
56	throw new Error('Invalid input');
57	}
58
59	// 2. Build prompt
60	const systemPrompt = `You are a ${context.role}.
61	Always respond in ${context.format}.
62	Tone: ${context.tone}`;
63
64	// 3. Call API
65	const response = await anthropic.messages.create({
66	model: 'claude-3-haiku-20240307',
67	max_tokens: 1000,
68	system: systemPrompt,
69	messages: [{
70	role: 'user',
71	content: userInput
72	}]
73	});
74
75	// 4. Parse and validate output
76	const output = response.content[0].text;
77	return parseOutput(output);
78	}
79	```
80
81	### Model Selection
82	\| Model \| Cost \| Speed \| Quality \| Use Case \|
83	\|-------\|------\|-------\|---------\|----------\|
84	\| GPT-4o \| $$$ \| Fast \| Best \| Complex tasks \|
85	\| GPT-4o-mini \| $ \| Fastest \| Good \| Most tasks \|
86	\| Claude 3.5 Sonnet \| $$ \| Fast \| Excellent \| Balanced \|
87	\| Claude 3 Haiku \| $ \| Fastest \| Good \| High volume \|
88	```
89
90	### Prompt Engineering for Products
91
92	Production-grade prompt design
93
94	When to use: When building AI product prompts
95
96	```javascript
97	## Prompt Engineering for Products
98
99	### Prompt Template Pattern
100	```javascript
101	const promptTemplates = {
102	emailWriter: {
103	system: `You are an expert email writer.
104	Write professional, concise emails.
105	Match the requested tone.
106	Never include placeholder text.`,
107	user: (input) => `Write an email:
108	Purpose: ${input.purpose}
109	Recipient: ${input.recipient}
110	Tone: ${input.tone}
111	Key points: ${input.points.join(', ')}
112	Length: ${input.length} sentences`,
113	},
114	};
115	```
116
117	### Output Control
118	```javascript
119	// Force structured output
120	const systemPrompt = `
121	Always respond with valid JSON in this format:
122	{
123	"title": "string",
124	"content": "string",
125	"suggestions": ["string"]
126	}
127	Never include any text outside the JSON.
128	`;
129
130	// Parse with fallback
131	function parseAIOutput(text) {
132	try {
133	return JSON.parse(text);
134	} catch {
135	// Fallback: extract JSON from response
136	const match = text.match(/\{[\s\S]*\}/);
137	if (match) return JSON.parse(match[0]);
138	throw new Error('Invalid AI output');
139	}
140	}
141	```
142
143	### Quality Control
144	\| Technique \| Purpose \|
145	\|-----------\|---------\|
146	\| Examples in prompt \| Guide output style \|
147	\| Output format spec \| Consistent structure \|
148	\| Validation \| Catch malformed responses \|
149	\| Retry logic \| Handle failures \|
150	\| Fallback models \| Reliability \|
151	```
152
153	### Cost Management
154
155	Controlling AI API costs
156
157	When to use: When building profitable AI products
158
159	```javascript
160	## AI Cost Management
161
162	### Token Economics
163	```javascript
164	// Track usage
165	async function callWithCostTracking(userId, prompt) {
166	const response = await anthropic.messages.create({...});
167
168	// Log usage
169	await db.usage.create({
170	userId,
171	inputTokens: response.usage.input_tokens,
172	outputTokens: response.usage.output_tokens,
173	cost: calculateCost(response.usage),
174	model: 'claude-3-haiku',
175	});
176
177	return response;
178	}
179
180	function calculateCost(usage) {
181	const rates = {
182	'claude-3-haiku': { input: 0.25, output: 1.25 }, // per 1M tokens
183	};
184	const rate = rates['claude-3-haiku'];
185	return (usage.input_tokens * rate.input +
186	usage.output_tokens * rate.output) / 1_000_000;
187	}
188	```
189
190	### Cost Reduction Strategies
191	\| Strategy \| Savings \|
192	\|----------\|---------\|
193	\| Use cheaper models \| 10-50x \|
194	\| Limit output tokens \| Variable \|
195	\| Cache common queries \| High \|
196	\| Batch similar requests \| Med

People who liked this also viewed

6 picks

Type