claude-flow 1.0.18 → 1.0.20

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "claude-flow",
-  "version": "1.0.18",
+  "version": "1.0.20",
   "description": "Advanced AI agent orchestration system for Claude Code",
   "main": "src/cli/main.ts",
   "bin": {

package/src/cli/commands/help.ts

@@ -177,6 +177,67 @@ const HELP_TOPICS: HelpTopic[] = [
     ],
     related: ['agents', 'workflows', 'coordination']
   },
+  {
+    name: 'claude',
+    description: 'Spawning Claude instances with specific configurations',
+    category: 'basic',
+    examples: [
+      {
+        description: 'Spawn Claude with web research capabilities',
+        command: 'claude-flow claude spawn "implement user authentication" --research --parallel',
+        explanation: 'Creates a Claude instance with WebFetchTool and BatchTool for parallel web research'
+      },
+      {
+        description: 'Spawn Claude without permission prompts',
+        command: 'claude-flow claude spawn "fix payment bug" --no-permissions',
+        explanation: 'Runs Claude with --dangerously-skip-permissions flag to avoid interruptions'
+      },
+      {
+        description: 'Spawn Claude with custom tools',
+        command: 'claude-flow claude spawn "analyze codebase" --tools "View,Edit,GrepTool,LS"',
+        explanation: 'Specifies exactly which tools Claude can use for the task'
+      },
+      {
+        description: 'Spawn Claude with test coverage target',
+        command: 'claude-flow claude spawn "write unit tests" --coverage 95 --commit feature',
+        explanation: 'Sets test coverage goal to 95% and commits after each feature'
+      },
+      {
+        description: 'Dry run to preview command',
+        command: 'claude-flow claude spawn "build API" --mode backend-only --dry-run',
+        explanation: 'Shows what would be executed without actually running Claude'
+      }
+    ],
+    tutorial: [
+      'The claude spawn command launches Claude instances with specific configurations.',
+      '',
+      'Available Options:',
+      '• --tools, -t: Specify allowed tools (default: View,Edit,Replace,GlobTool,GrepTool,LS,Bash)',
+      '• --no-permissions: Skip permission prompts with --dangerously-skip-permissions',
+      '• --config, -c: Path to MCP configuration file',
+      '• --mode, -m: Development mode (full, backend-only, frontend-only, api-only)',
+      '• --parallel: Enable BatchTool and dispatch_agent for parallel execution',
+      '• --research: Enable WebFetchTool for web research capabilities',
+      '• --coverage: Test coverage target percentage (default: 80)',
+      '• --commit: Commit frequency (phase, feature, manual)',
+      '• --verbose, -v: Enable verbose output',
+      '• --dry-run, -d: Preview what would be executed',
+      '',
+      'Environment Variables Set:',
+      '• CLAUDE_INSTANCE_ID: Unique identifier for the Claude instance',
+      '• CLAUDE_FLOW_MODE: Development mode setting',
+      '• CLAUDE_FLOW_COVERAGE: Target test coverage percentage',
+      '• CLAUDE_FLOW_COMMIT: Commit frequency setting',
+      '',
+      'Common Use Cases:',
+      '• Full-stack development: --mode full --parallel',
+      '• API development: --mode backend-only --coverage 90',
+      '• Bug fixing: --no-permissions --verbose',
+      '• Research tasks: --research --parallel',
+      '• Test writing: --coverage 95 --commit feature'
+    ],
+    related: ['agents', 'tasks', 'workflows']
+  },
   {
     name: 'workflows',
     description: 'Building complex multi-step workflows',

package/src/cli/simple-cli.js

@@ -38,6 +38,8 @@ EXAMPLES:
   claude-flow start                                    # Start orchestrator
   claude-flow agent spawn researcher --name "Bot"     # Spawn research agent
   claude-flow task create research "Analyze data"     # Create task
+  claude-flow claude spawn "implement auth" --research # Spawn Claude with web research
+  claude-flow claude spawn "fix bug" --no-permissions # Spawn Claude without permission prompts
   claude-flow config init                             # Initialize config
   claude-flow status                                  # Show system status
   claude-flow workflow my-workflow.json              # Execute workflow
@@ -274,7 +276,51 @@ async function main() {
             console.log(`📊 Coverage: ${flags.coverage || 80}%`);
             console.log(`💾 Commit: ${flags.commit || 'phase'}`);
             console.log('');
-            console.log('Note: Simple CLI is for demonstration. Use the full Claude-Flow CLI for actual execution.');
+            
+            // Build the actual claude command
+            const claudeArgs = [task];
+            claudeArgs.push('--allowedTools', tools);
+            
+            if (flags.noPermissions) {
+              claudeArgs.push('--dangerously-skip-permissions');
+            }
+            
+            if (flags.config) {
+              claudeArgs.push('--mcp-config', flags.config);
+            }
+            
+            if (flags.verbose) {
+              claudeArgs.push('--verbose');
+            }
+            
+            // Execute the actual claude command
+            try {
+              const command = new Deno.Command('claude', {
+                args: claudeArgs,
+                env: {
+                  ...Deno.env.toObject(),
+                  CLAUDE_INSTANCE_ID: instanceId,
+                  CLAUDE_FLOW_MODE: flags.mode || 'full',
+                  CLAUDE_FLOW_COVERAGE: (flags.coverage || 80).toString(),
+                  CLAUDE_FLOW_COMMIT: flags.commit || 'phase',
+                },
+                stdin: 'inherit',
+                stdout: 'inherit',
+                stderr: 'inherit',
+              });
+              
+              const child = command.spawn();
+              const status = await child.status;
+              
+              if (status.success) {
+                printSuccess(`Claude instance ${instanceId} completed successfully`);
+              } else {
+                printError(`Claude instance ${instanceId} exited with code ${status.code}`);
+              }
+            } catch (err) {
+              printError(`Failed to spawn Claude: ${err.message}`);
+              console.log('Make sure you have the Claude CLI installed.');
+            }
           }
           break;