codeswarm 0.1.0

package/orchestrate.sh

@@ -0,0 +1,467 @@
+#!/usr/bin/env bash
+#
+# orchestrate.sh — Multi-Agent Orchestration Script
+# Coordinates Claude Code, Gemini CLI, and Codex CLI in Plan → Execute → Review pipeline.
+#
+# Usage:
+#   ./orchestrate.sh --project <path> --task "description" [options]
+#
+# Options:
+#   --project <path>       Target project directory (required)
+#   --task <description>   Task description (required)
+#   --planner <agent>      Agent for planning   (claude|gemini|codex)  default: from config.yaml
+#   --executor <agent>     Agent for execution  (claude|gemini|codex)  default: from config.yaml
+#   --reviewer <agent>     Agent for review     (claude|gemini|codex)  default: from config.yaml
+#   --model <model>        Override model for planner (e.g. opus, sonnet, o3)
+#   --browser-test         Enable browser testing after review
+#   --test-url <url>       Base URL for browser tests
+#   --skip-plan            Skip planning phase (use existing plan.md)
+#   --skip-review          Skip review phase
+#   --config <path>        Custom config file (default: ./config.yaml)
+#   --verbose              Show full agent output
+#   --dry-run              Print commands without executing
+
+set -euo pipefail
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+CONFIG_FILE="${SCRIPT_DIR}/config.yaml"
+
+# ─── Colors ─────────────────────────────────────────────
+RED='\033[0;31m'
+GREEN='\033[0;32m'
+YELLOW='\033[1;33m'
+BLUE='\033[0;34m'
+CYAN='\033[0;36m'
+BOLD='\033[1m'
+NC='\033[0m'
+
+# ─── Defaults ───────────────────────────────────────────
+PROJECT=""
+TASK=""
+PLANNER=""
+EXECUTOR=""
+REVIEWER=""
+MODEL_OVERRIDE=""
+BROWSER_TEST=false
+TEST_URL=""
+SKIP_PLAN=false
+SKIP_REVIEW=false
+VERBOSE=false
+DRY_RUN=false
+
+# ─── Parse Arguments ────────────────────────────────────
+while [[ $# -gt 0 ]]; do
+  case $1 in
+    --project)     PROJECT="$2"; shift 2 ;;
+    --task)        TASK="$2"; shift 2 ;;
+    --planner)     PLANNER="$2"; shift 2 ;;
+    --executor)    EXECUTOR="$2"; shift 2 ;;
+    --reviewer)    REVIEWER="$2"; shift 2 ;;
+    --model)       MODEL_OVERRIDE="$2"; shift 2 ;;
+    --browser-test) BROWSER_TEST=true; shift ;;
+    --test-url)    TEST_URL="$2"; shift 2 ;;
+    --skip-plan)   SKIP_PLAN=true; shift ;;
+    --skip-review) SKIP_REVIEW=true; shift ;;
+    --config)      CONFIG_FILE="$2"; shift 2 ;;
+    --verbose)     VERBOSE=true; shift ;;
+    --dry-run)     DRY_RUN=true; shift ;;
+    -h|--help)     head -20 "$0" | tail -18; exit 0 ;;
+    *)             echo -e "${RED}Unknown option: $1${NC}"; exit 1 ;;
+  esac
+done
+
+# ─── Validate ────────────────────────────────────────────
+if [[ -z "$PROJECT" ]]; then
+  echo -e "${RED}Error: --project is required${NC}"
+  exit 1
+fi
+if [[ -z "$TASK" ]]; then
+  echo -e "${RED}Error: --task is required${NC}"
+  exit 1
+fi
+if [[ ! -d "$PROJECT" ]]; then
+  echo -e "${RED}Error: Project directory not found: $PROJECT${NC}"
+  exit 1
+fi
+
+# ─── Read Config (simple YAML parsing via grep/sed) ──────
+read_config() {
+  local key="$1"
+  local default="$2"
+  if [[ -f "$CONFIG_FILE" ]]; then
+    local val
+    val=$(grep -E "^\s+${key}:" "$CONFIG_FILE" 2>/dev/null | head -1 | sed 's/.*:\s*//' | sed 's/#.*//' | sed 's/"//g' | xargs)
+    if [[ -n "$val" ]]; then
+      echo "$val"
+      return
+    fi
+  fi
+  echo "$default"
+}
+
+# Apply config defaults where CLI flags not provided
+[[ -z "$PLANNER" ]]  && PLANNER=$(read_config "planner" "claude")
+[[ -z "$EXECUTOR" ]] && EXECUTOR=$(read_config "executor" "gemini")
+[[ -z "$REVIEWER" ]] && REVIEWER=$(read_config "reviewer" "codex")
+
+# ─── Artifacts & Logging ─────────────────────────────────
+ARTIFACTS="${PROJECT}/.codeswarm"
+SESSION_ID="run_$(date +%Y%m%d_%H%M%S)"
+SESSION_DIR="${ARTIFACTS}/sessions/${SESSION_ID}"
+PLAN_FILE="${SESSION_DIR}/plan.md"
+EXEC_LOG="${SESSION_DIR}/execution.log"
+REVIEW_FILE="${SESSION_DIR}/review.md"
+SCREENSHOTS_DIR="${SESSION_DIR}/screenshots"
+TIMESTAMP=$(date +"%Y-%m-%d_%H-%M-%S") # This variable is not used in the provided snippet, keeping it for consistency if used elsewhere.
+
+mkdir -p "$SESSION_DIR" "$SCREENSHOTS_DIR"
+
+# ─── Logging ─────────────────────────────────────────────
+log() { echo -e "${CYAN}[$(date +%H:%M:%S)]${NC} $1"; }
+success() { echo -e "${GREEN}✓${NC} $1"; }
+warn() { echo -e "${YELLOW}⚠${NC} $1"; }
+error() { echo -e "${RED}✗${NC} $1"; }
+header() {
+  echo ""
+  echo -e "${BOLD}━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━${NC}"
+  echo -e "${BOLD}  $1${NC}"
+  echo -e "${BOLD}━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━${NC}"
+}
+
+# ─── Agent Command Builders ─────────────────────────────
+run_agent() {
+  local agent="$1"
+  local prompt="$2"
+  local working_dir="$3"
+  local output_file="$4"
+  local permission_mode="${5:-plan}"
+
+  local cmd=""
+  case "$agent" in
+    claude)
+      local model_flag=""
+      if [[ -n "$MODEL_OVERRIDE" ]]; then
+        model_flag="--model $MODEL_OVERRIDE"
+      else
+        local default_model
+        default_model=$(read_config "claude" "opus")
+        [[ -n "$default_model" ]] && model_flag="--model $default_model"
+      fi
+      cmd="cd '$working_dir' && claude -p $model_flag --permission-mode $permission_mode '$prompt'"
+      ;;
+    gemini)
+      local model_flag=""
+      local default_model
+      default_model=$(read_config "gemini" "")
+      [[ -n "$default_model" ]] && model_flag="--model $default_model"
+      local approval_mode="default"
+      [[ "$permission_mode" == "plan" ]] && approval_mode="plan"
+      [[ "$permission_mode" == "auto_edit" ]] && approval_mode="auto_edit"
+      [[ "$permission_mode" == "bypassPermissions" ]] && approval_mode="yolo"
+      cmd="cd '$working_dir' && gemini -p '$prompt' $model_flag --approval-mode $approval_mode"
+      ;;
+    codex)
+      local sandbox_flag="--sandbox read-only"
+      [[ "$permission_mode" == "auto_edit" ]] && sandbox_flag=""
+      cmd="cd '$working_dir' && codex exec $sandbox_flag '$prompt'"
+      ;;
+    *)
+      error "Unknown agent: $agent"
+      exit 1
+      ;;
+  esac
+
+  if $DRY_RUN; then
+    echo -e "${YELLOW}[DRY RUN]${NC} $cmd"
+    echo "# Dry run — no output" > "$output_file"
+    return 0
+  fi
+
+  log "Running ${BOLD}$agent${NC}..."
+  if $VERBOSE; then
+    eval "$cmd" 2>&1 | tee "$output_file"
+  else
+    eval "$cmd" > "$output_file" 2>&1
+  fi
+  local exit_code=$?
+
+  if [[ $exit_code -ne 0 ]]; then
+    warn "$agent exited with code $exit_code"
+  else
+    success "$agent completed successfully"
+  fi
+  return $exit_code
+}
+
+# ─── Banner ──────────────────────────────────────────────
+echo ""
+echo -e "${BOLD}${CYAN}"
+echo "   ╔═══════════════════════════════════════════╗"
+echo "   ║     🤖 AGENTIC ORCHESTRATOR v1.0          ║"
+echo "   ╚═══════════════════════════════════════════╝"
+echo -e "${NC}"
+echo -e "  ${BOLD}Task:${NC}     $TASK"
+echo -e "  ${BOLD}Project:${NC}  $PROJECT"
+echo -e "  ${BOLD}Planner:${NC}  $PLANNER"
+echo -e "  ${BOLD}Executor:${NC} $EXECUTOR"
+echo -e "  ${BOLD}Reviewer:${NC} $REVIEWER"
+echo -e "  ${BOLD}Browser:${NC}  $BROWSER_TEST"
+echo ""
+
+# ─── PHASE 1: PLANNING ──────────────────────────────────
+if ! $SKIP_PLAN; then
+  header "📋 Phase 1: Planning ($PLANNER)"
+
+  PLAN_PROMPT="You are the PLANNER agent in a multi-agent development team.
+
+TASK: $TASK
+
+PROJECT DIRECTORY: $PROJECT
+
+Your job is to analyze the task and produce a detailed implementation plan.
+
+OUTPUT FORMAT (write as markdown):
+# Implementation Plan
+
+## Task Summary
+<One-paragraph summary of what needs to be done>
+
+## File Changes
+For each file to modify/create/delete:
+### [MODIFY|NEW|DELETE] <filepath>
+- What to change and why
+
+## Testing Strategy
+- How to verify the changes work
+- What to test manually
+- What automated tests to add
+
+## Risk Assessment
+- Potential issues or breaking changes
+- Dependencies to be aware of
+
+## Estimated Complexity
+Rate 1-10 and explain why.
+
+IMPORTANT:
+- Be specific about file paths relative to the project root
+- Include code snippets where helpful
+- Consider the existing codebase patterns
+- Think about edge cases"
+
+  PLANNER_PERMISSION=$(read_config "planner" "plan")
+  run_agent "$PLANNER" "$PLAN_PROMPT" "$PROJECT" "$PLAN_FILE" "$PLANNER_PERMISSION"
+
+  log "Plan saved to: $PLAN_FILE"
+  echo ""
+  echo -e "${YELLOW}─── Plan Preview ──────────────────────────────────${NC}"
+  head -30 "$PLAN_FILE"
+  echo -e "${YELLOW}───────────────────────────────────────────────────${NC}"
+  echo ""
+
+  # Optional: pause for human review
+  if [[ -t 0 ]]; then
+    read -p "$(echo -e ${BOLD})Review plan and press Enter to continue (or Ctrl+C to abort)...$(echo -e ${NC}) "
+  fi
+else
+  log "Skipping planning phase (using existing plan)"
+  if [[ ! -f "$PLAN_FILE" ]]; then
+    error "No existing plan found at $PLAN_FILE"
+    exit 1
+  fi
+fi
+
+# ─── PHASE 2: EXECUTION ─────────────────────────────────
+header "⚡ Phase 2: Execution ($EXECUTOR)"
+
+PLAN_CONTENT=$(cat "$PLAN_FILE")
+
+EXEC_PROMPT="You are the EXECUTOR agent in a multi-agent development team.
+
+TASK: $TASK
+
+You must implement the following plan EXACTLY. Do not deviate from the plan unless you encounter a blocker.
+
+=== PLAN ===
+$PLAN_CONTENT
+=== END PLAN ===
+
+INSTRUCTIONS:
+1. Read the relevant files first to understand current state
+2. Make the changes described in the plan
+3. After making changes, run any build/compile commands to verify
+4. If you encounter issues, fix them and document what you changed
+
+IMPORTANT:
+- Follow the plan precisely
+- Create new files where specified
+- Modify existing files as described
+- Run tests if the plan includes a testing strategy
+- Log any deviations from the plan"
+
+EXECUTOR_PERMISSION=$(read_config "executor" "auto_edit")
+run_agent "$EXECUTOR" "$EXEC_PROMPT" "$PROJECT" "$EXEC_LOG" "$EXECUTOR_PERMISSION"
+
+log "Execution log saved to: $EXEC_LOG"
+
+# ─── PHASE 3: REVIEW ────────────────────────────────────
+if ! $SKIP_REVIEW; then
+  header "🔍 Phase 3: Review ($REVIEWER)"
+
+  # Capture git diff for the reviewer
+  DIFF_OUTPUT=""
+  if command -v git &>/dev/null && git -C "$PROJECT" rev-parse --is-inside-work-tree &>/dev/null; then
+    DIFF_OUTPUT=$(git -C "$PROJECT" diff --stat 2>/dev/null || echo "No git diff available")
+    DIFF_DETAIL=$(git -C "$PROJECT" diff 2>/dev/null | head -500 || echo "")
+  fi
+
+  REVIEW_PROMPT="You are the REVIEWER agent in a multi-agent development team.
+
+TASK: $TASK
+
+You must review the changes made by the executor agent.
+
+=== ORIGINAL PLAN ===
+$PLAN_CONTENT
+=== END PLAN ===
+
+=== GIT DIFF SUMMARY ===
+$DIFF_OUTPUT
+=== END DIFF SUMMARY ===
+
+=== DIFF DETAIL (first 500 lines) ===
+$DIFF_DETAIL
+=== END DIFF ===
+
+REVIEW CHECKLIST:
+1. ✅ Does the implementation match the plan?
+2. ✅ Are there any bugs or logic errors?
+3. ✅ Code quality: naming, structure, patterns
+4. ✅ Are edge cases handled?
+5. ✅ Security considerations
+6. ✅ Performance concerns
+7. ✅ Missing tests or documentation
+8. ✅ Breaking changes
+
+OUTPUT FORMAT (markdown):
+# Code Review Report
+
+## Summary
+<Overall assessment: APPROVED | NEEDS_CHANGES | REJECTED>
+
+## Findings
+### Critical Issues
+- ...
+
+### Warnings
+- ...
+
+### Suggestions
+- ...
+
+## Plan Adherence
+<Did the executor follow the plan? Any deviations?>
+
+## Verdict
+<Final recommendation with reasoning>"
+
+  REVIEWER_PERMISSION=$(read_config "reviewer" "plan")
+  run_agent "$REVIEWER" "$REVIEW_PROMPT" "$PROJECT" "$REVIEW_FILE" "$REVIEWER_PERMISSION"
+
+  log "Review saved to: $REVIEW_FILE"
+fi
+
+# ─── PHASE 4: BROWSER TESTING ───────────────────────────
+if $BROWSER_TEST; then
+  header "🌐 Phase 4: Browser Testing (Playwright MCP)"
+
+  if [[ -z "$TEST_URL" ]]; then
+    TEST_URL=$(read_config "base_url" "http://localhost:4200")
+  fi
+
+  BROWSER_PROMPT="You are a QA agent performing browser-based testing.
+
+TASK: Test the changes made for: $TASK
+
+TEST URL: $TEST_URL
+
+INSTRUCTIONS:
+1. Navigate to $TEST_URL
+2. Login with test credentials if required
+3. Navigate to the relevant pages affected by the task
+4. Click buttons, fill forms, and interact with the UI
+5. Take a screenshot at EACH significant step
+6. Verify that the expected behavior matches the plan
+7. Report any visual glitches, broken layouts, or errors
+
+OUTPUT FORMAT (markdown):
+# Browser Test Report
+
+## Test Environment
+- URL: $TEST_URL
+- Browser: Chromium
+- Date: $TIMESTAMP
+
+## Test Steps
+### Step 1: [Description]
+- Action: ...
+- Expected: ...
+- Actual: ...
+- Screenshot: [filename]
+- Status: PASS | FAIL
+
+## Summary
+- Total steps: N
+- Passed: N
+- Failed: N
+- Screenshots: [list]"
+
+  # Use Claude with --chrome or Playwright MCP for browser testing
+  BROWSER_AGENT="${PLANNER}"  # default to planner agent for browser tests
+  TEST_REPORT="${AGENTIC_DIR}/test-report.md"
+
+  run_agent "$BROWSER_AGENT" "$BROWSER_PROMPT" "$PROJECT" "$TEST_REPORT" "default"
+
+  log "Test report saved to: $TEST_REPORT"
+fi
+
+# ─── FINAL REPORT ────────────────────────────────────────
+header "📊 Final Summary"
+
+REPORT_FILE="${AGENTIC_DIR}/report-${TIMESTAMP}.md"
+cat > "$REPORT_FILE" <<EOF
+# Orchestration Report
+
+**Date:** $TIMESTAMP
+**Task:** $TASK
+**Project:** $PROJECT
+
+## Agent Assignments
+| Role | Agent |
+|------|-------|
+| Planner | $PLANNER |
+| Executor | $EXECUTOR |
+| Reviewer | $REVIEWER |
+
+## Artifacts
+- Plan: [plan.md](./plan.md)
+- Execution Log: [execution.log](./execution.log)
+- Review: [review.md](./review.md)
+$(if $BROWSER_TEST; then echo "- Test Report: [test-report.md](./test-report.md)"; fi)
+
+## Quick Links
+- \`cat ${PLAN_FILE}\`
+- \`cat ${REVIEW_FILE}\`
+EOF
+
+success "Report saved to: $REPORT_FILE"
+echo ""
+echo -e "${GREEN}${BOLD}All phases complete!${NC}"
+echo -e "  📋 Plan:    ${PLAN_FILE}"
+echo -e "  ⚡ Log:     ${EXEC_LOG}"
+echo -e "  🔍 Review:  ${REVIEW_FILE}"
+if $BROWSER_TEST; then
+  echo -e "  🌐 Tests:   ${AGENTIC_DIR}/test-report.md"
+fi
+echo -e "  📊 Report:  ${REPORT_FILE}"
+echo ""

package/package.json

@@ -0,0 +1,62 @@
+{
+    "name": "codeswarm",
+    "version": "0.1.0",
+    "description": "Multi-agent AI orchestration framework — coordinate Claude, Gemini, Codex, Amp, and OpenCode as a unified dev team",
+    "license": "MIT",
+    "author": "Mustafa Kutlu <your-email@example.com>",
+    "repository": {
+        "type": "git",
+        "url": "https://github.com/mskutlu/codeswarm.git"
+    },
+    "homepage": "https://github.com/mskutlu/codeswarm#readme",
+    "bugs": "https://github.com/mskutlu/codeswarm/issues",
+    "keywords": [
+        "ai",
+        "agent",
+        "multi-agent",
+        "orchestration",
+        "claude",
+        "gemini",
+        "codex",
+        "amp",
+        "opencode",
+        "coding-agent",
+        "autonomous",
+        "prd",
+        "dashboard",
+        "workflow"
+    ],
+    "bin": {
+        "codeswarm": "bin/codeswarm.js"
+    },
+    "files": [
+        "bin/",
+        "coordinator.sh",
+        "orchestrate.sh",
+        "setup.sh",
+        "config.yaml",
+        "dashboard/",
+        ".codeswarm/",
+        "docs/",
+        "playwright.config.ts",
+        "COORDINATOR.md",
+        "AGENT_TIPS.md",
+        "TASK_PROTOCOL.md",
+        "BROWSER_TESTING.md",
+        "WORKFLOWS.md",
+        "LICENSE",
+        "README.md"
+    ],
+    "engines": {
+        "node": ">=18.0.0"
+    },
+    "scripts": {
+        "test": "bash coordinator.sh --project /tmp/codeswarm-test --prd docs/prd-example.md --dry-run",
+        "dashboard": "node dashboard/server.js"
+    },
+    "dependencies": {
+        "chokidar": "^3.6.0",
+        "express": "^4.21.0",
+        "ws": "^8.18.0"
+    }
+}

package/playwright.config.ts

@@ -0,0 +1,19 @@
+import { defineConfig } from '@playwright/test';
+
+export default defineConfig({
+    testDir: './tests',
+    timeout: 30000,
+    retries: 1,
+    use: {
+        baseURL: process.env.TEST_URL || 'http://localhost:4200',
+        headless: true,
+        screenshot: 'on',
+        video: 'retain-on-failure',
+        trace: 'on-first-retry',
+    },
+    reporter: [
+        ['html', { outputFolder: 'playwright-report' }],
+        ['json', { outputFile: '.codeswarm/test-results.json' }],
+    ],
+    outputDir: '.codeswarm/screenshots',
+});

package/setup.sh

@@ -0,0 +1,142 @@
+#!/usr/bin/env bash
+#
+# setup.sh — Install dependencies for the Agentic Orchestration Framework
+#
+
+set -euo pipefail
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+
+GREEN='\033[0;32m'
+YELLOW='\033[1;33m'
+RED='\033[0;31m'
+BOLD='\033[1m'
+NC='\033[0m'
+
+echo -e "${BOLD}🤖 Agentic Setup${NC}"
+echo ""
+
+# ─── Check Prerequisites ────────────────────────────────
+check_cmd() {
+  if command -v "$1" &>/dev/null; then
+    echo -e "  ${GREEN}✓${NC} $1 found: $(command -v "$1")"
+    return 0
+  else
+    echo -e "  ${RED}✗${NC} $1 not found"
+    return 1
+  fi
+}
+
+echo -e "${BOLD}Checking prerequisites...${NC}"
+MISSING=0
+check_cmd "claude"  || MISSING=1
+check_cmd "gemini"  || MISSING=1
+check_cmd "codex"   || MISSING=1
+check_cmd "node"    || MISSING=1
+check_cmd "npm"     || MISSING=1
+check_cmd "git"     || MISSING=1
+
+if [[ $MISSING -eq 1 ]]; then
+  echo ""
+  echo -e "${YELLOW}⚠ Some tools are missing. Install them before running the orchestrator.${NC}"
+  echo ""
+  echo "  Install CLI agents:"
+  echo "    Claude Code: npm install -g @anthropic/claude-code"
+  echo "    Gemini CLI:  npm install -g @anthropic/gemini-cli (or via Google's installer)"
+  echo "    Codex CLI:   npm install -g @openai/codex"
+  echo ""
+fi
+
+# ─── Install Node Dependencies ──────────────────────────
+echo ""
+echo -e "${BOLD}Installing Node dependencies...${NC}"
+cd "$SCRIPT_DIR"
+
+if [[ ! -f package.json ]]; then
+  npm init -y --silent
+fi
+
+npm install --save-dev playwright @playwright/test 2>/dev/null || true
+
+echo ""
+echo -e "${BOLD}Installing Playwright browsers...${NC}"
+npx playwright install chromium 2>/dev/null || true
+
+# ─── Make scripts executable ─────────────────────────────
+chmod +x "$SCRIPT_DIR/orchestrate.sh"
+chmod +x "$SCRIPT_DIR/setup.sh"
+
+# ─── MCP Configuration ──────────────────────────────────
+echo ""
+echo -e "${BOLD}MCP Configuration${NC}"
+echo ""
+echo "  To enable Playwright MCP for each agent, add the following:"
+echo ""
+echo -e "  ${YELLOW}Claude Code${NC} (~/.claude/mcp.json):"
+echo '    { "mcpServers": { "playwright": { "command": "npx", "args": ["@anthropic/mcp-server-playwright"] } } }'
+echo ""
+echo -e "  ${YELLOW}Gemini CLI${NC} (~/.gemini/settings.json):"
+echo '    { "mcpServers": { "playwright": { "command": "npx", "args": ["@anthropic/mcp-server-playwright"] } } }'
+echo ""
+echo -e "  ${YELLOW}Codex CLI${NC} (~/.codex/config.toml):"
+echo '    [mcp_servers.playwright]'
+echo '    command = "npx"'
+echo '    args = ["@anthropic/mcp-server-playwright"]'
+echo ""
+
+# ─── Create Example Test ─────────────────────────────────
+mkdir -p "$SCRIPT_DIR/tests"
+if [[ ! -f "$SCRIPT_DIR/tests/example.spec.ts" ]]; then
+  cat > "$SCRIPT_DIR/tests/example.spec.ts" << 'EOTEST'
+import { test, expect } from '@playwright/test';
+
+test.describe('Smoke Test', () => {
+  test('should load login page', async ({ page }) => {
+    await page.goto(process.env.TEST_URL || 'http://localhost:4200');
+    await page.screenshot({ path: '.codeswarm/screenshots/login-page.png' });
+    // Verify the page loads without errors
+    const title = await page.title();
+    expect(title).toBeTruthy();
+  });
+
+  test('should login and navigate', async ({ page }) => {
+    const baseUrl = process.env.TEST_URL || 'http://localhost:4200';
+    await page.goto(baseUrl);
+
+    // Attempt login (adjust selectors to your app)
+    const usernameInput = page.locator('input[type="text"], input[name="username"], #username');
+    const passwordInput = page.locator('input[type="password"], input[name="password"], #password');
+    const loginButton = page.locator('button[type="submit"], button:has-text("Login"), button:has-text("Giriş")');
+
+    if (await usernameInput.isVisible()) {
+      await usernameInput.fill(process.env.TEST_USER || 'admin');
+      await passwordInput.fill(process.env.TEST_PASS || 'admin');
+      await loginButton.click();
+      await page.waitForTimeout(3000);
+      await page.screenshot({ path: '.codeswarm/screenshots/after-login.png' });
+    }
+  });
+});
+EOTEST
+  echo -e "  ${GREEN}✓${NC} Created example test: tests/example.spec.ts"
+fi
+
+# ─── Create .gitignore ───────────────────────────────────
+if [[ ! -f "$SCRIPT_DIR/.gitignore" ]]; then
+  cat > "$SCRIPT_DIR/.gitignore" << 'EOIGNORE'
+node_modules/
+.codeswarm/
+*.log
+test-results/
+playwright-report/
+EOIGNORE
+  echo -e "  ${GREEN}✓${NC} Created .gitignore"
+fi
+
+echo ""
+echo -e "${GREEN}${BOLD}Setup complete!${NC}"
+echo ""
+echo "  Next steps:"
+echo "    1. Review config.yaml and adjust defaults"
+echo "    2. Run: ./orchestrate.sh --project <path> --task \"your task\""
+echo ""