claude-flow-novice 2.18.13 → 2.18.14

package/.claude/skills/cfn-cerebras-coordinator/lib/tdd-conversation-coordinator.ts

@@ -0,0 +1,569 @@
+/**
+ * TDD Conversation Coordinator (TypeScript)
+ *
+ * Iterative TDD with Cerebras SDK + conversation memory for self-correction.
+ * Reads context file contents and embeds them in prompts.
+ *
+ * Uses GLM 4.6 best practices:
+ * - disable_reasoning: true (skip verbose reasoning for code generation)
+ * - temperature: 0.6 (instruction following)
+ * - top_p: 0.95
+ */
+
+import { readFileSync, writeFileSync, mkdirSync, existsSync } from 'fs';
+import { dirname, basename, extname, join } from 'path';
+import { execSync } from 'child_process';
+import { fileURLToPath } from 'url';
+import Cerebras from '@cerebras/cerebras_cloud_sdk';
+
+// ESM compatibility: Create __dirname equivalent
+const __filename = fileURLToPath(import.meta.url);
+const __dirname = dirname(__filename);
+
+// Configuration
+const CEREBRAS_MODEL = process.env.CEREBRAS_MODEL || 'zai-glm-4.6';
+
+// Initialize Cerebras client (uses CEREBRAS_API_KEY env var by default)
+const client = new Cerebras({
+  warmTCPConnection: true, // Reduces TTFT
+});
+
+// Types
+export interface TDDCoordinatorOptions {
+  agentId: string;
+  feature: string;
+  filePath: string;
+  contextFiles?: string[];
+  testCommand: string;
+  maxIterations?: number;
+  verbose?: boolean;
+}
+
+interface ConversationMessage {
+  role: 'system' | 'user' | 'assistant';
+  content: string;
+  timestamp?: string;
+}
+
+export interface TDDResult {
+  success: boolean;
+  implementationFile: string;
+  testFile: string;
+  iterations: number;
+  conversationId: string;
+  error?: string;
+}
+
+
+export class TDDConversationCoordinator {
+  private options: Required<Omit<TDDCoordinatorOptions, 'contextFiles'>> & { contextFiles: string[] };
+  private conversation: ConversationMessage[] = [];
+  private conversationId: string;
+
+  constructor(options: TDDCoordinatorOptions) {
+    this.options = {
+      maxIterations: 5,
+      verbose: false,
+      contextFiles: [],
+      ...options
+    };
+    this.conversationId = `${options.agentId}-${Date.now()}`;
+    this.initConversation();
+  }
+
+  private initConversation(): void {
+    this.conversation = [{
+      role: 'system',
+      content: `You are a TDD expert. Follow Red-Green-Refactor strictly. Use Given/When/Then test structure. Generate ONLY code - no markdown code blocks, no explanations, just raw code.`
+    }];
+  }
+
+  private log(message: string): void {
+    if (this.options.verbose) {
+      console.error(`[${new Date().toISOString()}] ${message}`);
+    }
+  }
+
+  private logError(message: string): void {
+    console.error(`[ERROR] ${message}`);
+  }
+
+  /**
+   * Call Cerebras API with full conversation history
+   */
+  private async callCerebras(prompt: string): Promise<string> {
+    // Add user message to conversation
+    this.conversation.push({
+      role: 'user',
+      content: prompt,
+      timestamp: new Date().toISOString()
+    });
+
+    this.log(`Calling Cerebras (${CEREBRAS_MODEL}) with ${this.conversation.length} messages...`);
+
+    const completion = await client.chat.completions.create({
+      model: CEREBRAS_MODEL,
+      messages: this.conversation.map(m => ({ role: m.role, content: m.content })),
+      max_completion_tokens: 20000,
+      temperature: 0.6,  // Recommended for instruction following
+      top_p: 0.95,       // Recommended default
+      // @ts-expect-error disable_reasoning is undocumented but supported
+      disable_reasoning: true,  // Skip verbose reasoning for code generation
+    });
+
+    const content = completion.choices?.[0]?.message?.content;
+    if (!content) {
+      throw new Error('Empty response from Cerebras API');
+    }
+
+    // Add assistant response to conversation
+    this.conversation.push({
+      role: 'assistant',
+      content,
+      timestamp: new Date().toISOString()
+    });
+
+    this.log(`Received response (${content.length} chars)`);
+    return content;
+  }
+
+  /**
+   * Read context files and return their contents formatted for prompts
+   */
+  private readContextFiles(): string {
+    if (!this.options.contextFiles.length) {
+      return '';
+    }
+
+    const sections: string[] = [];
+
+    for (const filePath of this.options.contextFiles) {
+      const trimmedPath = filePath.trim();
+      if (!trimmedPath) continue;
+
+      if (!existsSync(trimmedPath)) {
+        this.logError(`Context file not found: ${trimmedPath}`);
+        continue;
+      }
+
+      try {
+        const content = readFileSync(trimmedPath, 'utf-8');
+        this.log(`Loaded context: ${trimmedPath} (${content.length} chars)`);
+        sections.push(`## ${trimmedPath}\n\`\`\`\n${content}\n\`\`\``);
+      } catch (err) {
+        this.logError(`Failed to read ${trimmedPath}: ${err}`);
+      }
+    }
+
+    return sections.join('\n\n');
+  }
+
+  /**
+   * Determine test file path based on implementation file
+   */
+  private getTestFilePath(): string {
+    const ext = extname(this.options.filePath);
+    const base = basename(this.options.filePath, ext);
+    const dir = dirname(this.options.filePath);
+
+    switch (ext) {
+      case '.ts':
+      case '.tsx':
+        return join(dir, `${base}.test.ts`);
+      case '.js':
+      case '.jsx':
+        return join(dir, `${base}.test.js`);
+      case '.py':
+        return join(dir, `test_${base}.py`);
+      case '.go':
+        return join(dir, `${base}_test.go`);
+      case '.rs':
+        return join(dir, `${base}_test.rs`);
+      default:
+        return join(dir, `${base}.test${ext}`);
+    }
+  }
+
+  /**
+   * Extract code from response (removes markdown blocks if present)
+   */
+  private extractCode(content: string): string {
+    // Try to extract from markdown code block
+    const codeBlockMatch = content.match(/```(?:\w+)?\n([\s\S]*?)```/);
+    if (codeBlockMatch) {
+      return codeBlockMatch[1].trim();
+    }
+    // Return as-is if no code block
+    return content.trim();
+  }
+
+  /**
+   * Run test command and capture output
+   */
+  private runTests(): { success: boolean; output: string } {
+    this.log(`Running: ${this.options.testCommand}`);
+
+    try {
+      // Change to implementation file's directory for test execution
+      // This fixes vitest/jest issues with absolute paths
+      const workDir = dirname(this.options.filePath);
+      this.log(`Working directory: ${workDir}`);
+
+      const output = execSync(this.options.testCommand, {
+        encoding: 'utf-8',
+        stdio: ['pipe', 'pipe', 'pipe'],
+        timeout: 60000, // 60 second timeout
+        cwd: workDir // Run from the directory containing the files
+      });
+      return { success: true, output };
+    } catch (err: unknown) {
+      const error = err as { stderr?: string; stdout?: string; message?: string };
+      const output = error.stderr || error.stdout || error.message || 'Unknown error';
+      return { success: false, output };
+    }
+  }
+
+  /**
+   * Save conversation to file for learning
+   */
+  private saveConversation(): void {
+    try {
+      const dir = join(__dirname, '..', 'conversations');
+      mkdirSync(dir, { recursive: true });
+
+      const date = new Date().toISOString().split('T')[0];
+      const file = join(dir, `${date}-${this.options.agentId}.json`);
+
+      const data = {
+        conversationId: this.conversationId,
+        metadata: {
+          agentId: this.options.agentId,
+          feature: this.options.feature,
+          filePath: this.options.filePath,
+          contextFiles: this.options.contextFiles,
+          completedAt: new Date().toISOString()
+        },
+        conversation: this.conversation
+      };
+
+      writeFileSync(file, JSON.stringify(data, null, 2));
+      this.log(`Conversation saved: ${file}`);
+    } catch (err) {
+      this.logError(`Failed to save conversation: ${err}`);
+    }
+  }
+
+  /**
+   * Run the full TDD workflow
+   */
+  async run(): Promise<TDDResult> {
+    console.log('🧪 TDD Conversation Coordinator (TypeScript)');
+    console.log('='.repeat(50));
+    console.log(`Feature:        ${this.options.feature}`);
+    console.log(`Target File:    ${this.options.filePath}`);
+    console.log(`Agent ID:       ${this.options.agentId}`);
+    console.log(`Max Iterations: ${this.options.maxIterations}`);
+    console.log(`Test Command:   ${this.options.testCommand}`);
+    console.log('='.repeat(50));
+    console.log();
+
+    const testFile = this.getTestFilePath();
+    const contextContent = this.readContextFiles();
+
+    // === PHASE 2: Generate tests (RED) ===
+    console.log('🔴 Phase 2 (RED): Generating failing tests...');
+
+    const testPrompt = `# TDD Red Phase: Write Failing Tests
+
+## Feature
+${this.options.feature}
+
+## Target Implementation File
+${this.options.filePath}
+
+## Context Files
+${contextContent || '(none provided)'}
+
+## Requirements
+1. Write comprehensive tests using STANDARD test framework syntax (do NOT create custom helper functions):
+   - **TypeScript/JavaScript**: Use describe/it/expect blocks from vitest or jest
+     - Import framework: \`import { describe, it, expect } from 'vitest';\`
+     - Import functions: \`import { functionName } from './filename';\`
+     - Structure: \`describe('feature', () => { it('should...', () => { expect(...).toBe(...) }) })\`
+   - **Python**: Use pytest with def test_* functions
+     - Import: \`from filename import ClassName\` or \`import pytest\`
+     - Structure: \`def test_feature(): assert value == expected\`
+   - **Rust**: Use #[test] attribute or #[cfg(test)] module
+     - Import: \`use super::*;\` or \`use crate::module::*;\`
+     - Structure: \`#[test] fn test_name() { assert_eq!(result, expected) }\`
+2. Tests MUST fail initially (no implementation exists yet)
+3. Cover happy path AND edge cases
+4. Include setup/teardown if needed
+5. **CRITICAL: Include ALL necessary import statements at the top of the test file**
+6. **Do NOT create custom test helper functions (Given/When/Then helpers, etc.)**
+
+IMPORTANT: Generate ONLY the test file content. No markdown code blocks, no explanations, just raw code.`;
+
+    let testCode: string;
+    try {
+      testCode = this.extractCode(await this.callCerebras(testPrompt));
+    } catch (err) {
+      return {
+        success: false,
+        implementationFile: this.options.filePath,
+        testFile,
+        iterations: 0,
+        conversationId: this.conversationId,
+        error: `Test generation failed: ${err}`
+      };
+    }
+
+    // Write test file
+    mkdirSync(dirname(testFile), { recursive: true });
+    writeFileSync(testFile, testCode);
+    console.log(`📝 Tests written to: ${testFile}`);
+
+    // === PHASE 4: Generate implementation (GREEN) ===
+    console.log('🟢 Phase 4 (GREEN): Generating implementation...');
+
+    const implPrompt = `# TDD Green Phase: Write Implementation
+
+The tests you just wrote are now failing. Write the MINIMUM implementation to make them pass.
+
+## Test File Content
+\`\`\`
+${testCode}
+\`\`\`
+
+## Target Implementation File
+${this.options.filePath}
+
+## Requirements
+1. Implement ONLY what the tests require
+2. Don't add extra features (YAGNI)
+3. Make all tests pass
+4. Follow existing code patterns from context
+5. **CRITICAL: Export all functions/classes that the tests import**
+6. **For TypeScript/JavaScript: use \`export function name()\` or \`export { name }\`**
+7. **For Python: functions are exported by default**
+8. **For Rust: use \`pub fn name()\` for public functions**
+
+IMPORTANT: Generate ONLY the implementation file content. No markdown code blocks, no explanations, just raw code.`;
+
+    let implCode: string;
+    try {
+      implCode = this.extractCode(await this.callCerebras(implPrompt));
+    } catch (err) {
+      return {
+        success: false,
+        implementationFile: this.options.filePath,
+        testFile,
+        iterations: 0,
+        conversationId: this.conversationId,
+        error: `Implementation generation failed: ${err}`
+      };
+    }
+
+    // Write implementation file
+    mkdirSync(dirname(this.options.filePath), { recursive: true });
+    writeFileSync(this.options.filePath, implCode);
+    console.log(`📝 Implementation written to: ${this.options.filePath}`);
+
+    // === PHASE 5-6: Test + Fix Loop ===
+    for (let iteration = 1; iteration <= this.options.maxIterations; iteration++) {
+      console.log();
+      console.log(`--- Iteration ${iteration}/${this.options.maxIterations} ---`);
+
+      const { success, output } = this.runTests();
+
+      if (success) {
+        console.log('🎉 SUCCESS! All tests pass.');
+        this.saveConversation();
+
+        return {
+          success: true,
+          implementationFile: this.options.filePath,
+          testFile,
+          iterations: iteration,
+          conversationId: this.conversationId
+        };
+      }
+
+      console.log('❌ Tests failed, sending error to Cerebras for fix...');
+
+      // Detect if error is test-related or implementation-related
+      const errorLower = output.toLowerCase();
+
+      // Implementation issues (assertion failures, logic errors) - check FIRST
+      const isImplementationIssue =
+        errorLower.includes('expected') ||          // Assertion: expected X but got Y
+        errorLower.includes('assertion') ||         // Assertion failure
+        errorLower.includes('to be') ||             // expect(x).toBe(y) failures
+        errorLower.includes('to equal') ||          // expect(x).toEqual(y) failures
+        errorLower.includes('received') ||          // Jest: Expected X, Received Y
+        errorLower.includes('assert') ||            // assert failures
+        /\d+ (passing|failed)/.test(errorLower);    // Test results with failures = impl issue
+
+      // Test file issues (syntax, imports, missing files) - only if NOT impl issue
+      const isTestFileIssue = !isImplementationIssue && (
+        errorLower.includes('no test') ||
+        errorLower.includes('cannot find test') ||
+        errorLower.includes('test file') ||
+        errorLower.includes('error: no tests') ||
+        errorLower.includes('cannot find module') ||     // Import errors
+        errorLower.includes('is not defined') ||         // Missing imports
+        errorLower.includes('syntaxerror') ||            // Syntax errors in test
+        (errorLower.includes('cannot find') && testFile && errorLower.includes(basename(testFile)))
+      );
+
+      const targetFile = isTestFileIssue ? testFile : this.options.filePath;
+      const fileType = isTestFileIssue ? 'TEST' : 'IMPLEMENTATION';
+
+      console.log(`🔍 Detected ${fileType} file issue, will fix: ${targetFile}`);
+
+      const fixPrompt = `# Fix Required - Tests Still Failing (Iteration ${iteration})
+
+## Error Output
+\`\`\`
+${output.slice(0, 2000)}
+\`\`\`
+
+## Target File to Fix
+${fileType} file: ${targetFile}
+
+## Instructions
+1. Carefully analyze the error output above
+2. Review your previous ${fileType.toLowerCase()} in the conversation
+3. Identify what's wrong (logic error, **missing imports**, missing edge case, incorrect behavior)
+4. **CRITICAL: If error says "X is not defined" or "cannot find name X":**
+   - Check if X exists in the implementation file
+   - If YES: Add import statement to the TEST file (not implementation)
+   - Example: \`import { X } from './filename';\`
+5. Fix ONLY the specific issue in the ${fileType} file
+6. Preserve all working functionality
+
+IMPORTANT: Generate the COMPLETE fixed ${fileType} file content for ${basename(targetFile)}. No markdown code blocks, no explanations, just raw code.`;
+
+      try {
+        const fixedCode = this.extractCode(await this.callCerebras(fixPrompt));
+        writeFileSync(targetFile, fixedCode);
+        console.log(`📝 Fixed ${fileType} written to: ${targetFile}`);
+      } catch (err) {
+        this.logError(`Fix generation failed: ${err}`);
+      }
+    }
+
+    console.log();
+    console.log(`❌ Failed after ${this.options.maxIterations} iterations`);
+    this.saveConversation();
+
+    return {
+      success: false,
+      implementationFile: this.options.filePath,
+      testFile,
+      iterations: this.options.maxIterations,
+      conversationId: this.conversationId,
+      error: `Failed after ${this.options.maxIterations} iterations`
+    };
+  }
+}
+
+// === CLI Entry Point ===
+async function main(): Promise<void> {
+  const args = process.argv.slice(2);
+
+  if (args.includes('--help') || args.includes('-h')) {
+    console.log(`
+TDD Conversation Coordinator (TypeScript)
+
+Usage:
+  npx ts-node tdd-conversation-coordinator.ts [options]
+
+Required Options:
+  --agent-id ID           Agent identifier
+  --feature DESCRIPTION   Feature to implement
+  --file-path PATH        Target implementation file
+  --test-command CMD      Command to run tests
+
+Optional:
+  --context-files FILES   Comma-separated context file paths
+  --max-iterations N      Max fix iterations (default: 5)
+  --verbose               Enable verbose logging
+
+Environment Variables:
+  CEREBRAS_API_KEY       Cerebras API key (required, auto-loaded by SDK)
+  CEREBRAS_MODEL         Model name (default: zai-glm-4.6)
+
+Example:
+  npx ts-node tdd-conversation-coordinator.ts \\
+    --agent-id tdd-001 \\
+    --feature "Email validator" \\
+    --file-path ./src/email.ts \\
+    --test-command "npm test email.test.ts" \\
+    --context-files "./src/types.ts,./src/utils.ts" \\
+    --max-iterations 3 \\
+    --verbose
+`);
+    process.exit(0);
+  }
+
+  // Parse arguments
+  const options: Partial<TDDCoordinatorOptions> = {};
+
+  for (let i = 0; i < args.length; i++) {
+    const arg = args[i];
+
+    if (arg === '--verbose') {
+      options.verbose = true;
+      continue;
+    }
+
+    if (arg.startsWith('--') && i + 1 < args.length) {
+      const key = arg.slice(2).replace(/-([a-z])/g, (_, c) => c.toUpperCase());
+      const value = args[++i];
+
+      switch (key) {
+        case 'agentId':
+          options.agentId = value;
+          break;
+        case 'feature':
+          options.feature = value;
+          break;
+        case 'filePath':
+          options.filePath = value;
+          break;
+        case 'testCommand':
+          options.testCommand = value;
+          break;
+        case 'contextFiles':
+          options.contextFiles = value.split(',').map(f => f.trim());
+          break;
+        case 'maxIterations':
+          options.maxIterations = parseInt(value, 10);
+          break;
+      }
+    }
+  }
+
+  // Validate required options
+  if (!options.agentId || !options.feature || !options.filePath || !options.testCommand) {
+    console.error('Error: Missing required arguments');
+    console.error('Run with --help for usage information');
+    process.exit(1);
+  }
+
+  // Run coordinator
+  const coordinator = new TDDConversationCoordinator(options as TDDCoordinatorOptions);
+  const result = await coordinator.run();
+
+  console.log();
+  console.log('Result:');
+  console.log(JSON.stringify(result, null, 2));
+
+  process.exit(result.success ? 0 : 1);
+}
+
+main().catch(err => {
+  console.error('Fatal error:', err);
+  process.exit(1);
+});