@measey/mycoder-agent/dist/core/toolAgent/config.js

Agent module for mycoder - an AI-powered software development assistant

import * as fs from 'fs';
import * as os from 'os';
import * as path from 'path';
/**
 * Default configuration for the tool agent
 */
export const DEFAULT_CONFIG = {
    maxIterations: 200,
    getSystemPrompt: getDefaultSystemPrompt,
};
/**
 * Gets the default system prompt with contextual information about the environment
 */
export function getDefaultSystemPrompt(toolContext) {
    // Gather context using Node.js APIs for cross-platform compatibility
    const getCurrentDirectory = () => {
        try {
            return process.cwd();
        }
        catch (error) {
            return `[Error getting current directory: ${error.message}]`;
        }
    };
    const getFileList = () => {
        try {
            const currentDir = process.cwd();
            const files = fs.readdirSync(currentDir, { withFileTypes: true });
            // Sort directories first, then files
            files.sort((a, b) => {
                if (a.isDirectory() && !b.isDirectory())
                    return -1;
                if (!a.isDirectory() && b.isDirectory())
                    return 1;
                return a.name.localeCompare(b.name);
            });
            // Format similar to 'ls -la' output
            const fileDetails = files.map((file) => {
                try {
                    const stats = fs.statSync(path.join(currentDir, file.name));
                    const permissions = getPermissionString(stats);
                    const links = stats.nlink || 1;
                    const owner = process.platform === 'win32' ? 'OWNER' : stats.uid.toString();
                    const group = process.platform === 'win32' ? 'GROUP' : stats.gid.toString();
                    const size = stats.size;
                    const mtime = stats.mtime
                        .toDateString()
                        .split(' ')
                        .slice(1)
                        .join(' ');
                    const type = file.isDirectory() ? 'd' : '-';
                    return `${type}${permissions} ${links.toString().padStart(2)} ${owner.padEnd(10)} ${group.padEnd(10)} ${size.toString().padStart(8)} ${mtime} ${file.name}${file.isDirectory() ? '/' : ''}`;
                }
                catch (err) {
                    return `[Error getting details for ${file.name}: ${err.message}]`;
                }
            });
            return `total ${files.length}\n${fileDetails.join('\n')}`;
        }
        catch (error) {
            return `[Error getting file listing: ${error.message}]`;
        }
    };
    const getPermissionString = (stats) => {
        const mode = stats.mode;
        const permissions = [
            mode & 0o400 ? 'r' : '-',
            mode & 0o200 ? 'w' : '-',
            mode & 0o100 ? 'x' : '-',
            mode & 0o040 ? 'r' : '-',
            mode & 0o020 ? 'w' : '-',
            mode & 0o010 ? 'x' : '-',
            mode & 0o004 ? 'r' : '-',
            mode & 0o002 ? 'w' : '-',
            mode & 0o001 ? 'x' : '-',
        ].join('');
        return permissions;
    };
    const getSystemInfo = () => {
        try {
            return `${os.type()} ${os.hostname()} ${os.release()} ${os.platform()} ${os.arch()}`;
        }
        catch (error) {
            return `[Error getting system information: ${error.message}]`;
        }
    };
    const context = {
        pwd: getCurrentDirectory(),
        files: getFileList(),
        system: getSystemInfo(),
        datetime: new Date().toString(),
        githubMode: toolContext.githubMode,
    };
    const githubModeInstructions = context.githubMode
        ? [
            '',
            '## GitHub Mode',
            'GitHub mode is enabled. You should work with GitHub issues and PRs as part of your workflow:',
            '- Start from existing GitHub issues (gh issue view 21) or create new ones (gh issue create) for tasks',
            "- Create branches for issues you're working on",
            '- Make commits with descriptive messages',
            '- Create PRs when work is complete (gh pr create)',
            '- Create additional GitHub issues for follow-up tasks or ideas',
            '',
            'You should use the Github CLI tool, gh, and the git cli tool, git, that you can access via shell commands.',
            '',
            'When creating GitHub issues, PRs, or comments via the gh cli tool, use the shellStart or shellExecute stdinContent parameter for multiline content:',
            '- Use the stdinContent parameter to pass the content directly to the command',
            '- For example: `shellStart({ command: "gh issue create --body-stdin", stdinContent: "Issue description here with **markdown** support", description: "Creating a new issue" })`',
            '- This approach preserves formatting, newlines, and special characters correctly without requiring temporary files',
        ].join('\n')
        : '';
    return [
        'You are an AI agent that can use tools to accomplish tasks.',
        '',
        'Current Context:',
        `Directory: ${context.pwd}`,
        'Files:',
        context.files,
        `System: ${context.system}`,
        `DateTime: ${context.datetime}`,
        githubModeInstructions,
        '',
        '## Resource Management',
        'You will receive periodic status updates showing your token usage and active background tasks.',
        'If your token usage approaches 50% of the maximum, you should use the compactHistory tool to reduce context size.',
        'The compactHistory tool will summarize older messages while preserving recent context.',
        'Status updates are sent every 5 iterations and also whenever token usage exceeds 50% of the maximum.',
        '',
        'You prefer to call tools in parallel when possible because it leads to faster execution and less resource usage.',
        'When done, call the agentDone tool with your results to indicate that the sequence has completed.',
        '',
        'For coding tasks:',
        '0. Try to break large tasks into smaller sub-tasks that can be completed and verified sequentially.',
        "   - trying to make lots of changes in one go can make it really hard to identify when something doesn't work",
        '   - use sub-agents for each sub-task, leaving the main agent in a supervisory role',
        '   - when possible ensure the project compiles/builds and the tests pass after each sub-task',
        '   - give the sub-agents the guidance and context necessary be successful',
        '1. First understand the context by:',
        '   - Reading README.md, CONTRIBUTING.md, and similar documentation',
        '   - Checking project configuration files (e.g., package.json)',
        '   - Understanding coding standards',
        '2. Ensure changes:',
        '   - Follow project conventions',
        '   - Build successfully',
        '   - Pass all tests',
        '3. Update documentation as needed',
        '4. Consider adding documentation if you encountered setup/understanding challenges',
        '',
        'Feel free to use AI friendly search engines via the browser tools to search for information or for ideas when you get stuck.',
        '',
        'When you run into issues or unexpected results, take a step back and read the project documentation and configuration files and look at other source files in the project for examples of what works.',
        '',
        'Use sub-agents for parallel tasks, providing them with specific context they need rather than having them rediscover it.',
        (() => {
            if (!toolContext.customPrompt)
                return '';
            if (typeof toolContext.customPrompt === 'string') {
                return `\n\n${toolContext.customPrompt}`;
            }
            // It's an array of strings
            return `\n\n${toolContext.customPrompt.join('\n')}`;
        })(),
    ].join('\n');
}
//# sourceMappingURL=config.js.map