jqai 0.1.2

Generate and execute jq programs using LLMs

jqai

Always forgetting the jq syntax? Wish you had a magic box that translated your desires to the right syntax? Look no further than jqai!

Generate and execute jq programs using LLMs. Transform JSON data using natural language descriptions instead of learning complex jq syntax.

Usage

cat data.json | uvx jqai "get all user names and emails"

Prerequisites

• jq: Must be installed on your system. Install jq
• API Key: OpenAI API key or Echo API key

Setting up API Key

Option 1: OpenAI API Key

export OPENAI_API_KEY="your-openai-api-key"

Option 2: Echo API (Recommended)

Get an Echo API key at echo.merit.systems:

export OPENAI_API_KEY="echo_your-echo-api-key"

Usage

jqai works in two modes:

1. Piped JSON Mode

Transform JSON data by piping it to jqai:

cat data.json | jqai "get all user names and emails"

curl -s https://api.github.com/users/octocat | jqai "extract login, name, and public repo count"

2. Command Generation Mode

Generate complete commands that fetch and process data:

jqai "show the latest 5 issues from simonw/datasette repo"

jqai "get weather data for San Francisco and extract temperature"

Examples

Basic JSON Transformation

Given a JSON file users.json:

{
  "users": [
    {"name": "Alice", "age": 30, "city": "New York"},
    {"name": "Bob", "age": 25, "city": "San Francisco"},
    {"name": "Charlie", "age": 35, "city": "New York"}
  ]
}

Extract names and cities:

cat users.json | jqai "get names and cities of all users"

Filter users from New York:

cat users.json | jqai "show only users from New York"

Count users by city:

cat users.json | jqai "count users by city"

API Data Processing

Process GitHub API data:

curl -s https://api.github.com/repos/microsoft/vscode/issues | jqai "show issue titles and their authors"

Get repository statistics:

curl -s https://api.github.com/repos/python/cpython | jqai "extract name, stars, forks, and primary language"

Complex Transformations

Group and aggregate data:

cat sales.json | jqai "group sales by product and sum the amounts"

Nested object manipulation:

cat complex.json | jqai "flatten the nested user profiles and extract email domains"

Command Line Options

Option	Short	Description	Default
--model	-m	OpenAI model to use	gpt-4.1-mini
--preview-length	-p	Bytes of input for LLM context	1024
--output-only	-o	Show generated program only	False
--silent	-s	Reduce output verbosity	True
--verbose	-v	Show prompts and responses	False

Examples with Options

Generate program without executing:

cat data.json | jqai "extract user emails" --output-only

Use a different model:

cat data.json | jqai "complex analysis" --model gpt-4-turbo-preview

Verbose mode for debugging:

cat data.json | jqai "transform data" --verbose

Advanced Usage

Learning jq

Use --output-only to see the generated jq programs and learn:

$ cat users.json | jqai "get names of users over 30" --output-only
.users[] | select(.age > 30) | .name

Custom Preview Length

For large JSON files, adjust the preview length:

cat large-file.json | jqai "analyze data structure" --preview-length 2048

Combining with Other Tools

Chain with other command-line tools:

jqai "get latest commits from torvalds/linux" | head -10

cat logs.json | jqai "extract error messages" | grep -i "critical"

How It Works

1. Input Analysis: jqai analyzes your natural language description
2. Context Building: For piped data, it reads a preview to understand the JSON structure
3. Program Generation: Uses LLM to generate an appropriate jq program or shell command
4. Execution: Runs the generated program with your data

Piped Mode Flow

JSON Data → Preview → LLM → jq Program → Execute → Output

Command Mode Flow

Description → LLM → Shell Command → Execute → Output

Error Handling

jqai handles common errors gracefully:

• Missing API Key: Clear instructions on how to set up authentication
• jq Errors: Passes through jq error messages for debugging
• Network Issues: Proper error reporting for API calls
• Broken Pipes: Handles interrupted output streams

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License.

Troubleshooting

Common Issues

"jq: command not found"

• Install jq: brew install jq (macOS) or apt-get install jq (Ubuntu)

"OPENAI_API_KEY environment variable is not set"

• Set your API key: export OPENAI_API_KEY="your-key"
• For Echo API, get a key at echo.merit.systems

Generated program doesn't work as expected

• Use --verbose to see the generated program
• Try --output-only to inspect the jq syntax
• Provide more specific descriptions in your natural language query

Getting Help

For more complex queries, try being more specific:

Instead of: "process the data" Try: "extract user names and count them by department"

Instead of: "fix the JSON" Try: "remove null values and flatten nested objects"