askai

README

Ask AI

A tool for interacting with an AI service from the command line.

ltheisen@MM269697-PC ~
$ askai complete --user "how do i use logs in scriptrunner for jira" --max-tokens 1000 --stream
1. Import the `Logger` class:
   ```
   import com.onresolve.scriptrunner.runner.util.Logger
   ```

2. Create a logger instance:
   ```
   def log = Logger.log(this.class.name)
   ```

3. Log at different levels using the following methods:
   - `log.debug()`: for debug messages
   - `log.info()`: for informational messages
   - `log.warn()`: for warning messages
   - `log.error()`: for error messages

Example:
   ```
   log.debug('Debug message')
   log.info('Info message')
   log.warn('Warning message')
   log.error('Error message')
   ```

4. Output of the log messages is displayed in the ScriptRunner logs:
   Jira → Issues and filters → ScriptRunner → Logs

Installation

askai is a self contained binary that has pre-built releases for various platforms. You may find this script valuable for installation:

# note this command uses clconf which can be found here:
#   https://github.com/pastdev/clconf
(
  # where do you want this installed?
  binary="${HOME}/.local/bin/askai"
  # one of linux, darwin, windows
  platform="linux"
  curl \
    --location \
    --output "${binary}" \
    "$(
      curl --silent https://api.github.com/repos/pastdev/askai/releases/latest \
        | clconf \
          --pipe \
          jsonpath "$..assets[*][?(@.name =~ /askai-${platform/windows/windows.exe}/)].browser_download_url" \
          --first)"
  chmod 0755 "${binary}"
)

The ai alias

You may also find it useful to add this alias to one of your shell profile scripts:

alias ai='askai complete --user '

Configuration

Configuration is loaded by default from the following directories (in order):

• /etc/askai.d
• ~/.config/askai.d
• ./askai.d

The format of configuration is:

default_endpoint: windows_ollama
endpoints:
  # An API endpoint with both chat completion and image generation models
  grok:
    api_type: OPEN_AI
    auth_token: <OMITTED>
    base_url: "https://api.x.ai/v1"
    # optionally supply defaults for requests
    chat_completion_defaults:
      messages:
      - content: |
          provide concise responses with minimal ceremony. do not use a
          conversational style.
        role: system
      model: grok-3-latest
    image_defaults:
      model: grok-2-image-latest
  windows_ollama:
    api_type: OPEN_AI
    base_url: "http://172.22.144.1:11434/v1"
    empty_messages_limit: 300
    # optionally supply defaults for requests
    chat_completion_defaults:
      max_tokens: 250
      messages:
      - content: |
          provide concise responses with minimal ceremony. do not use a
          conversational style.
        role: system
      model: mistral

More options are available, see the Config type in the askai package for details.

Using Ollama

Start ollama on windows. You need to specify 0.0.0.0 or it will only listen on 127.0.0.1.

PS C:\Users\lucas> $env:OLLAMA_HOST="0.0.0.0:11434"
PS C:\Users\lucas> Get-NetTCPConnection -LocalPort 11434

LocalAddress                        LocalPort RemoteAddress                       RemotePort State       AppliedSetting OwningProcess
------------                        --------- -------------                       ---------- -----       -------------- -------------
127.0.0.1                           11434     0.0.0.0                             0          Listen                     10276


PS C:\Users\lucas> ollama serve

Then locate the IP address and associated with this service from WSL:

ltheisen@ltserver ~/git/pastdev/askai
$ grep ^nameserver /etc/resolv.conf
nameserver 172.22.144.1

This information will be needed for your configuration.

Development

This project has a snapshot script that makes using a snapshot build version of the source easy. To make use of this feature, create a symlink to the script somewhere in your path. For example:

# assuming you are currently in the base directory of this project
ln -s "$(pwd)/askais ~/.local/bin/askais

If you like the ai alias, you can updated it to point to this script:

alias ai="askais complete --user"