HARPA AI

https://harpa.ai/guides
Admin
HARPA AI is a practical web AI agent that integrates with multiple LLMs to summarize content,...
Tokens:35,562
Snippets:340
Trust Score:9.7
Update:9 months ago
Show doc for...
Tokens:
Raw
### Example GPT Prompt for Data Extraction

Source: https://harpa.ai/guides/connecting-harpa-to-make-com

This prompt instructs a GPT agent to act as a Data Research AI, extract specific product details (name, description, price, URL) from a webpage, and format the output as a JSON array. It specifies response constraints and provides an example JSON structure.

```Prompt
You are a Data Research AI Agent.

Your task is to extract data from a web page and respond with JSON text as requested. 

Follow these instructions:
- Analyze the product web page and find its name, price, and formulate a brief description of the product (maximum 200 characters).
- Provide the price as a number without currency symbols.
- Respond in {{language}}. 
- Be concise and to the point.
- Respond with a JSON array containing objects with 4 fields: 1 - product, 2 - description, 3 - price, 4 - url.
- Write nothing other than the JSON.
- Wrap your JSON response in a code block.

Example JSON response:
{
	"product": "Product name",
	"description": "Brief product description",
	"price": 1000,
	"url": "page url"
}


Product URL: {{url}}

Product webpage data: {{page}}

JSON response:

```

--------------------------------

### Start Ollama Server for HARPA AI (Windows Command Prompt)

Source: https://harpa.ai/guides/connecting-harpa-ollama-local-llms

Command to start the Ollama server on Windows, allowing connections from HARPA AI by setting the OLLAMA_ORIGINS environment variable using the 'set' command.

```cmd
set OLLAMA_ORIGINS=chrome-extension://* && ollama serve
```

--------------------------------

### OpenAI Function Specification Example

Source: https://harpa.ai/guides/function-calling

This JSON object demonstrates the structure for declaring a function to be used with OpenAI's function calling feature. It includes the function name, a description, and a detailed schema for its parameters, specifying types, descriptions, and required fields.

```json
{
  "name": "get_current_weather",
  "description": "Get the current weather",
  "parameters": {
    "type": "object",
    "properties": {
      "location": {
        "type": "string",
        "description": "The city and state, e.g. San Francisco, CA"
      },
      "format": {
        "type": "string",
        "enum": ["celsius", "fahrenheit"],
        "description": "The temperature unit to use. Infer this from the users location."
      }
    },
    "required": ["location", "format"]
  }
}
```

--------------------------------

### Creating Custom AI Commands with HARPA AI

Source: https://harpa.ai/guides/getting-started

Demonstrates how to create custom AI commands in HARPA AI by defining prompts that utilize web page content. This allows for tailored AI interactions based on the current page.

```plaintext
Please summarize this web page: {{page}}
```

--------------------------------

### Start Ollama Server for HARPA AI (Mac Terminal)

Source: https://harpa.ai/guides/connecting-harpa-ollama-local-llms

Command to start the Ollama server on macOS, allowing connections from HARPA AI by setting the OLLAMA_ORIGINS environment variable.

```bash
OLLAMA_ORIGINS=chrome-extension://* ollama serve
```

--------------------------------

### Example of a Bad Prompt: Too General

Source: https://harpa.ai/guides/custom-page-aware-chatgpt-commands

Illustrates a common pitfall in prompt engineering: using overly general or ambiguous language. This type of prompt, 'Tell me something interesting,' provides no specific direction and is unlikely to yield a useful or targeted response from an AI.

```text
Tell me something interesting

```

--------------------------------

### Custom Customer Service Agent Command in HARPA AI

Source: https://harpa.ai/guides/page-parameter-page-crawler

An example of creating a custom HARPA AI command to act as a Customer Service Agent. It involves steps to get user query and website, extract relevant links using SERP, and then use GPT to generate a response based on fetched page content.

```HARPA AI
/
CREATE
ASK -> query
ASK -> website
CALC: SET PARAMETER -> data = {{serp {{query}} site:{{website}}}} 
CALC: EXTRACT SERP LINKS -> links = {{data}}
GPT -> response = "Please ignore all previous instructions.\nThe output should be in {{language}}.\n\nAs a Customer Service Agent and the keeper of the database for the website \"{{website}}\",\n\nI want you to create a detailed response to my query, adhering to a\nspecific format I provide:\n- Use only the [INFORMATION] that I provide to you.\n- Do not use any other tools or sources.\n- Please integrate insights from the additional information into your\n  response and ensure the inclusion of links in a markdown format: [[1]](URL).\n- Please avoid general phrases and unrelated words in your response to my question.\n- Do not fabricate facts or information.\n\nFOLLOW THE FORMAT:\n## Customer Service Agent Response to \"{{query}}\"\n... detailed response text with links, e.g., [[1]](URL).\n\n[USER QUESTION]: {{query}}\n\n[INFORMATION]:\n\n- {{links.0.title}} {{links.0.url}} {{links.0.description}}\n{{page {{query}} {{links.0.url}} limit=20%}}\n\n- {{links.1.title}} {{links.1.url}} {{links.1.description}}\n{{page {{query}} {{links.1.url}} limit=10%}}\n\n- {{links.2.title}} {{links.2.url}} {{links.2.description}}\n{{page {{query}} {{links.2.url}} limit=10%}}\n\n[Your detailed response to the user's query]:\n"
```

--------------------------------

### Function Parameter Type Example (String)

Source: https://harpa.ai/guides/function-calling

This snippet illustrates how to define a string parameter within a function's specification for OpenAI's function calling. It shows the 'type' as 'string' and includes a 'description' to guide the AI on its purpose and expected format.

```json
"location": {
  "type": "string",
  "description": "The city and state, e.g. San Francisco, CA",
}
```

--------------------------------

### Improving a Basic Prompt for FAQ Generation

Source: https://harpa.ai/guides/custom-page-aware-chatgpt-commands

This snippet demonstrates how to transform a vague prompt into a detailed and effective one for generating a FAQ about a webpage. It specifies the desired output format (markdown), content structure (subheadings, questions, answers), tone, and word count, while incorporating the webpage content.

```text
Write me FAQ about {{page}}

```

```text
I want you to write a FAQ about a product described in the web page. FAQ should include at least five subheadings that includes a question and an answer. For every question provide a concise answer. Style output with markdown. Use a friendly tone. Write approximately 750 words in total.

Web page content: {{page}}

FAQ:

```

--------------------------------

### Prompt for Math Teacher Role-Play Explanation

Source: https://harpa.ai/guides/custom-page-aware-chatgpt-commands

This prompt sets up the AI to act as a math teacher, explaining mathematical concepts, equations, or problems in an easy-to-understand manner. It allows for step-by-step instructions, technique demonstrations, and suggestions for further study resources.

```text
I want you to act as a math teacher. I will give you a mathematical equation, concept or a problem, and your job will be to explain them in easy-to-understand terms to me. This could include providing step-by-step instructions for solving a problem, demonstrating various techniques or suggesting online resources for further study.

The equation / concept / problem: {{page}}

Explanation:

```

--------------------------------

### Function Parameter Type Example (Enum)

Source: https://harpa.ai/guides/function-calling

This example shows how to define an enumerated string parameter for function calling. The 'enum' keyword restricts the possible values to a predefined list (e.g., 'celsius', 'fahrenheit'), ensuring the AI provides valid input for the specified unit.

```json
"format": {
  "type": "string",
  "enum": ["celsius", "fahrenheit"],
  "description": "The temperature unit to use. Infer this from the users location.",
}
```

--------------------------------

### Concatenate GPT Answers for Summarization (HARPA AI)

Source: https://harpa.ai/guides/chunking

This HARPA AI command example shows how to combine results from multiple chunked GPT responses into a single summary. It uses the {{gpt}} parameter to reference the concatenated output from previous steps, creating a final summary of the entire document.

```HARPA AI Command
I would like you to analyze the text and write a TL;DR summary, followed by the 5 most important takeaways from the text.
Begin with a TL;DR section that is 30 words long.

Output format:
## Wiki Page TL;DR:
brief text summary

## Key Takeaways:
1️⃣ .....
2️⃣ .....
3️⃣ .....

TEXT: {{gpt}}

Summary:
```

--------------------------------

### Prompt for Domain Name Generation

Source: https://harpa.ai/guides/custom-page-aware-chatgpt-commands

This prompt instructs the AI to function as a domain name generator. Given company or product information, it aims to provide a list of available .com domain names that are 10 characters or less, incorporating relevant SEO keywords.

```text
I want you to act as a domain name generator. I'll tell you my company name or product description and your goal is to provide a list of free domain names. Reply with the list of domain names and nothing else. The domain names should be 10 letters maximum. The domain names should include SEO keywords for the product or the business whenever possible. The domain names should end in .com.

Information about the company: {{p1}}

Domain names:

```

--------------------------------

### Create Custom Command for Wikipedia Summary with Chunking (HARPA AI)

Source: https://harpa.ai/guides/chunking

This example demonstrates how to create a custom HARPA AI command to summarize a Wikipedia article. It uses the {{chunk}} parameter to process parts of the article and formats the output with a TL;DR and numbered emoji bullet points.

```HARPA AI Command
Ignore all previous instructions. I want you to only answer in {{language}}.

Please extract the key takeaways from the following Wikipedia article.
Provide a summary without echoing the text itself back to me.
The summary should include a minimum of three bullet points and a 30 words TL;DR section at the beginning.
Label each bullet point with a numbered emoji.

Output format:
## TL;DR
brief text summary

## Key Takeaways:
1️⃣ .....
2️⃣ .....
3️⃣ .....

TEXT: {{chunk}}
```
Tokens:
Raw