Providers/GitHub.ps1
|
<#
.SYNOPSIS Invokes the GitHub Models API to generate responses using specified models. .DESCRIPTION The Invoke-GitHubProvider function sends requests to the GitHub Models API and returns the generated content. It requires a GitHub token to be set in the environment variable 'GITHUB_TOKEN'. .PARAMETER ModelName The name of the GitHub model to use (e.g., 'gpt-4o-mini'). Note: Model availability might change. .PARAMETER Messages An array of hashtables containing the messages to send to the model. .EXAMPLE $Message = New-ChatMessage -Prompt 'Explain quantum computing in simple terms' $response = Invoke-GitHubProvider -ModelName 'gpt-4o-mini' -Messages $Message .NOTES Requires the GITHUB_TOKEN environment variable to be set with a valid GitHub token with appropriate permissions. The API endpoint used is 'https://models.inference.ai.azure.com/chat/completions'. This might be subject to change. #> function Invoke-GitHubProvider { [CmdletBinding()] param( [Parameter(Mandatory)] [string]$ModelName, [Parameter(Mandatory)] [hashtable[]]$Messages ) if (-not $env:GITHUB_TOKEN) { Write-Error "Please set the GITHUB_TOKEN environment variable with a valid GitHub token." return } $headers = @{ "Content-Type" = "application/json" "Authorization" = "Bearer $($env:GITHUB_TOKEN)" } $body = @{ "messages" = $Messages "model" = $ModelName } # Note: This endpoint might change. Refer to official GitHub documentation if issues arise. $Uri = "https://models.inference.ai.azure.com/chat/completions" $params = @{ Uri = $Uri Method = 'POST' Headers = $headers Body = $body | ConvertTo-Json -Depth 10 } try { $response = Invoke-RestMethod @params # Assuming the response structure is similar to OpenAI's return $response.choices[0].message.content } catch { $statusCode = $_.Exception.Response.StatusCode.value__ $errorMessage = $_.ErrorDetails.Message Write-Error "GitHub Models API Error (HTTP $statusCode): $errorMessage" return "Error calling GitHub Models API: $($_.Exception.Message)" } } |