ask

func AskExpert ¶

func AskExpert(ctx context.Context, input string) (reply string, err error)

AskExpert calls the AI expert model for consultation and returns the reply.

This function executes the dscli chat command via internal/shell, writes the input content to a temporary file, and passes it to dscli via the --input flag, avoiding command-line length limits and stdin issues.

Parameters:

ctx: context object for passing execution environment configuration
input: input text to send to the AI model, can be any length
       (limited by system memory)

Returns:

reply: the AI model's response text. Returns empty string if execution
       fails without obtaining a reply.
err: error during execution. Returns nil on success. Common errors include:
     - dscli command execution failure
     - temporary file creation/write failure

Details:

The function invokes the AI model by executing the following command:
     dscli chat --no-color --no-timestamp --histsize 0 --model <model_name> --input <temp_file>
where the model name is specified by ModelDeepseekReasoner.

Notes:

• Ensure the dscli CLI tool is properly installed and configured.
• Input content is passed via a temporary file that is automatically cleaned up after execution.

Example:

ctx := context.Background()
reply, err := AskExpert(ctx, "Please analyze the quality of this code")
if err != nil {
    log.Printf("Consultation failed: %v", err)
} else {
    fmt.Println(reply)
}

See also:

• shell.SimpleExecute: function for executing shell commands
• handleAskExpert: tool handler function that uses this function

func AskExpertWithRole ¶ added in v0.7.8

func AskExpertWithRole(ctx context.Context, input, role string) (reply string, err error)

AskExpertWithRole calls the AI model for consultation with a specified role (dev/expert/review).

This function executes the dscli chat command via internal/shell, writes the input content to a temporary file, and passes it to dscli via the --input and --role flags.

Parameters:

ctx: context object
input: input text to send to the AI model
role: role (dev/expert/review)

Returns:

reply: the AI model's response text
err: error during execution