improvements and promptfoo

rubrics/code-quality.yaml

@@ -0,0 +1,32 @@
+name: code-quality
+description: Evaluates generated code for quality and maintainability
+passingThreshold: 0.7
+
+criteria:
+  - name: readability
+    weight: 0.3
+    description: Code is easy to read and understand
+    examples:
+      good: "Clear variable names, logical flow, proper indentation"
+      bad: "Single-letter variables, deeply nested logic, inconsistent style"
+  
+  - name: correctness
+    weight: 0.4
+    description: Code correctly implements the intended behavior
+    examples:
+      good: "Handles edge cases, correct algorithm, proper error handling"
+      bad: "Missing edge cases, off-by-one errors, swallowed exceptions"
+  
+  - name: efficiency
+    weight: 0.2
+    description: Code uses appropriate data structures and algorithms
+    examples:
+      good: "O(n) where O(n) is optimal, avoids unnecessary allocations"
+      bad: "O(n²) when O(n) is possible, creates objects in tight loops"
+  
+  - name: maintainability
+    weight: 0.1
+    description: Code is easy to modify and extend
+    examples:
+      good: "Single responsibility, low coupling, clear interfaces"
+      bad: "God functions, tight coupling, magic numbers"

rubrics/documentation.yaml

@@ -0,0 +1,32 @@
+name: documentation
+description: Evaluates quality of code documentation and docstrings
+passingThreshold: 0.65
+
+criteria:
+  - name: completeness
+    weight: 0.35
+    description: Documentation covers all parameters, return values, and exceptions
+    examples:
+      good: "Fully documents args, returns, raises, and includes usage example"
+      bad: "Missing parameter descriptions or return type"
+  
+  - name: accuracy
+    weight: 0.35
+    description: Documentation accurately describes the function's behavior
+    examples:
+      good: "Description matches implementation, types are correct"
+      bad: "Outdated docs that don't match current behavior"
+  
+  - name: examples
+    weight: 0.2
+    description: Includes helpful usage examples
+    examples:
+      good: "Shows common use cases with expected outputs"
+      bad: "No examples or only trivial ones"
+  
+  - name: style
+    weight: 0.1
+    description: Follows project/language documentation conventions
+    examples:
+      good: "Uses standard docstring format (Google, NumPy, or reStructuredText)"
+      bad: "Inconsistent or non-standard format"

rubrics/error-messages.yaml

@@ -0,0 +1,25 @@
+name: error-messages
+description: Evaluates quality of error messages
+passingThreshold: 0.6
+
+criteria:
+  - name: clarity
+    weight: 0.4
+    description: Error message clearly explains what went wrong
+    examples:
+      good: "Invalid email format: 'not-an-email' is missing '@' symbol"
+      bad: "Error: validation failed"
+  
+  - name: actionability
+    weight: 0.4
+    description: Error message suggests how to fix the problem
+    examples:
+      good: "File not found. Create the file or check the path spelling."
+      bad: "ENOENT"
+  
+  - name: context
+    weight: 0.2
+    description: Error message includes relevant context (file, line, values)
+    examples:
+      good: "TypeError at line 42 in auth.py: expected str, got int (value=123)"
+      bad: "type error"

src/cli/commands/pipeline.ts

@@ -0,0 +1,257 @@
+import { Command } from 'commander';
+import { existsSync, readFileSync, writeFileSync, mkdirSync } from 'fs';
+import { join, resolve } from 'path';
+import { analyze } from '../../introspector/index.js';
+import { generateEvalSpec, generateEvalSpecInteractive } from '../../analyzer/index.js';
+import { renderSpec, detectFramework as detectRenderFramework } from '../../renderers/index.js';
+import { runTests, formatResults, DEFAULT_SANDBOX_CONFIG } from '../../runners/index.js';
+import { createTracer, saveTrace } from '../../observability/index.js';
+import { generatePromptfooConfig, generateTestProvider } from '../../promptfoo/index.js';
+import type { EvalSpec } from '../../analyzer/types.js';
+
+const EVALUCLAUDE_DIR = '.evaluclaude';
+
+interface PipelineOptions {
+  output?: string;
+  interactive?: boolean;
+  focus?: string;
+  maxScenarios: string;
+  testDir: string;
+  framework?: string;
+  skipAnalyze?: boolean;
+  skipRender?: boolean;
+  skipRun?: boolean;
+  promptfoo?: boolean;
+  quiet?: boolean;
+}
+
+export const pipelineCommand = new Command('pipeline')
+  .description('Run the full eval generation pipeline: introspect → analyze → render → run')
+  .argument('[path]', 'Path to the repository to analyze', '.')
+  .option('-o, --output <dir>', 'Output directory for all artifacts', '.evaluclaude')
+  .option('-i, --interactive', 'Enable interactive mode with clarifying questions')
+  .option('--focus <modules>', 'Comma-separated list of modules/functions to focus on')
+  .option('--max-scenarios <n>', 'Maximum number of test scenarios to generate', '10')
+  .option('--test-dir <dir>', 'Directory for generated tests', './tests/generated')
+  .option('-f, --framework <framework>', 'Test framework (pytest, vitest, jest)')
+  .option('--skip-analyze', 'Skip analysis, use existing spec')
+  .option('--skip-render', 'Skip rendering, use existing tests')
+  .option('--skip-run', 'Skip test execution')
+  .option('--promptfoo', 'Generate Promptfoo configuration for UI viewing')
+  .option('--quiet', 'Suppress progress messages')
+  .action(async (repoPath: string, options: PipelineOptions) => {
+    const absolutePath = resolve(repoPath);
+    const log = options.quiet ? () => {} : console.log;
+    const outputDir = options.output || EVALUCLAUDE_DIR;
+
+    console.log('\n🚀 Evaluclaude Pipeline');
+    console.log('═'.repeat(50));
+    console.log(`   Repository: ${absolutePath}`);
+    console.log(`   Output: ${outputDir}`);
+    console.log('═'.repeat(50) + '\n');
+
+    // Ensure output directories exist
+    mkdirSync(outputDir, { recursive: true });
+    mkdirSync(options.testDir, { recursive: true });
+
+    const specPath = join(outputDir, 'spec.json');
+    const tracesDir = join(outputDir, 'traces');
+    const resultsDir = join(outputDir, 'results');
+
+    mkdirSync(tracesDir, { recursive: true });
+    mkdirSync(resultsDir, { recursive: true });
+
+    let spec: EvalSpec;
+
+    // Step 1: Introspection + Analysis
+    if (options.skipAnalyze && existsSync(specPath)) {
+      log('📋 Using existing EvalSpec...');
+      spec = JSON.parse(readFileSync(specPath, 'utf-8'));
+      log(`   Loaded: ${specPath} (${spec.scenarios.length} scenarios)\n`);
+    } else {
+      log('🔬 Step 1: Introspecting codebase...');
+      
+      try {
+        const repoSummary = await analyze({
+          root: absolutePath,
+          onProgress: options.quiet ? undefined : (msg) => log(`   ${msg}`),
+        });
+
+        log(`   Files: ${repoSummary.files.length}`);
+        log(`   Languages: ${repoSummary.languages.join(', ')}`);
+        log('');
+
+        log('🤖 Step 2: Generating EvalSpec with Claude...\n');
+
+        const focus = options.focus?.split(',').map(s => s.trim());
+        const maxScenarios = parseInt(options.maxScenarios, 10);
+
+        let result;
+        if (options.interactive) {
+          const { default: inquirer } = await import('inquirer');
+          
+          result = await generateEvalSpecInteractive(
+            repoSummary,
+            async (question: string) => {
+              const { answer } = await inquirer.prompt([{
+                type: 'input',
+                name: 'answer',
+                message: `🤖 Claude asks: ${question}`,
+              }]);
+              return answer;
+            },
+            { focus, maxScenarios }
+          );
+        } else {
+          result = await generateEvalSpec(repoSummary, {
+            interactive: false,
+            focus,
+            maxScenarios,
+          });
+        }
+
+        spec = result.spec;
+
+        // Save the spec
+        writeFileSync(specPath, JSON.stringify(spec, null, 2));
+
+        log(`\n✅ EvalSpec generated!`);
+        log(`   Scenarios: ${spec.scenarios.length}`);
+        log(`   Tokens: ${result.tokensUsed}`);
+        log(`   Saved: ${specPath}\n`);
+      } catch (error) {
+        console.error('\n❌ Analysis failed:', error instanceof Error ? error.message : error);
+        process.exit(1);
+      }
+    }
+
+    // Step 2: Render tests
+    if (!options.skipRender) {
+      log('📝 Step 3: Rendering test files...');
+
+      try {
+        const framework = (options.framework as 'pytest' | 'vitest' | 'jest') || detectRenderFramework(spec);
+        
+        const renderResult = await renderSpec(spec, {
+          outputDir: options.testDir,
+          framework,
+          includeFixtures: true,
+          generateMocks: true,
+          dryRun: false,
+        });
+
+        log(`   Framework: ${framework}`);
+        log(`   Files: ${renderResult.stats.fileCount}`);
+        log(`   Scenarios: ${renderResult.stats.scenarioCount}`);
+        log(`   Assertions: ${renderResult.stats.assertionCount}`);
+        log(`   Output: ${options.testDir}\n`);
+      } catch (error) {
+        console.error('\n❌ Rendering failed:', error instanceof Error ? error.message : error);
+        process.exit(1);
+      }
+    }
+
+    // Step 3: Run tests
+    if (!options.skipRun) {
+      log('🧪 Step 4: Running tests...\n');
+
+      try {
+        const framework = (options.framework as 'pytest' | 'vitest' | 'jest') || detectRenderFramework(spec);
+        const tracer = createTracer(spec.repo.name);
+
+        tracer.recordIntrospection({
+          filesAnalyzed: spec.scenarios.map(s => s.target.module),
+          totalFunctions: spec.scenarios.length,
+          duration: 0,
+        });
+
+        tracer.recordGeneration({
+          scenariosGenerated: spec.scenarios.length,
+          filesWritten: [options.testDir],
+        });
+
+        const result = await runTests(
+          options.testDir,
+          {
+            framework,
+            sandbox: true,
+            timeout: 300000,
+            parallel: false,
+            cwd: process.cwd(),
+          },
+          DEFAULT_SANDBOX_CONFIG
+        );
+
+        tracer.recordExecution({
+          testsPassed: result.summary.passed,
+          testsFailed: result.summary.failed,
+          testsSkipped: result.summary.skipped,
+        });
+
+        for (const test of result.tests) {
+          if (test.status === 'failed' || test.status === 'error') {
+            tracer.recordTestFailure({
+              scenarioId: test.id,
+              testName: test.name,
+              error: test.error?.message || 'Unknown error',
+              stack: test.error?.stack,
+            });
+          }
+        }
+
+        const trace = tracer.finalize();
+        const tracePath = await saveTrace(trace);
+
+        log(formatResults(result));
+        log(`📊 Trace saved: ${tracePath}`);
+        log(`   View with: evaluclaude view ${trace.id}\n`);
+
+        // Save results
+        const resultsPath = join(resultsDir, `run-${Date.now()}.json`);
+        writeFileSync(resultsPath, JSON.stringify(result, null, 2));
+
+      } catch (error) {
+        console.error('\n❌ Test execution failed:', error instanceof Error ? error.message : error);
+        process.exit(1);
+      }
+    }
+
+    // Step 4: Generate Promptfoo config
+    if (options.promptfoo) {
+      log('📦 Step 5: Generating Promptfoo configuration...');
+
+      try {
+        const configPath = join(outputDir, 'promptfooconfig.yaml');
+        const providerPath = join(outputDir, 'providers', 'test-runner.py');
+        const framework = (options.framework as 'pytest' | 'vitest' | 'jest') || detectRenderFramework(spec);
+
+        await generatePromptfooConfig(spec, {
+          testDir: options.testDir,
+          outputPath: configPath,
+          framework,
+          includeTraceLinks: true,
+          providerPath,
+        });
+
+        await generateTestProvider(providerPath);
+
+        log(`   Config: ${configPath}`);
+        log(`   Provider: ${providerPath}`);
+        log(`\n   Launch UI with: evaluclaude ui\n`);
+      } catch (error) {
+        console.error('\n❌ Promptfoo config generation failed:', error instanceof Error ? error.message : error);
+      }
+    }
+
+    console.log('═'.repeat(50));
+    console.log('✅ Pipeline complete!');
+    console.log('═'.repeat(50));
+    console.log(`\nNext steps:`);
+    console.log(`   View traces:     evaluclaude view --last`);
+    console.log(`   List all traces: evaluclaude traces`);
+    if (options.promptfoo) {
+      console.log(`   Launch UI:       evaluclaude ui`);
+      console.log(`   Run Promptfoo:   evaluclaude eval --spec ${specPath}`);
+    }
+    console.log('');
+  });

src/cli/commands/run.ts

@@ -10,6 +10,7 @@ import {
   DEFAULT_SANDBOX_CONFIG 
 } from '../../runners/index.js';
 import { createTracer, saveTrace } from '../../observability/index.js';
+import { exportToPromptfooFormat } from '../../promptfoo/results-exporter.js';
 import type { EvalSpec } from '../../analyzer/types.js';
 
 export const runCommand = new Command('run')
@@ -25,6 +26,7 @@ export const runCommand = new Command('run')
   .option('-o, --output <file>', 'Output results to JSON file')
   .option('--trace', 'Record execution trace', true)
   .option('--no-trace', 'Disable execution tracing')
+  .option('--export-promptfoo', 'Export results in Promptfoo format', false)
   .option('-w, --watch', 'Watch mode (rerun on changes)', false)
   .action(async (testDir: string, options) => {
     try {
@@ -109,6 +111,16 @@ export const runCommand = new Command('run')
         console.log(`\n📁 Results saved to: ${options.output}`);
       }
 
+      // Export to Promptfoo format for UI viewing
+      if (options.exportPromptfoo) {
+        const exportPath = await exportToPromptfooFormat(result, spec, {
+          outputDir: '.evaluclaude/results',
+          evalId: `eval-${Date.now()}`,
+        });
+        console.log(`\n📦 Promptfoo results exported: ${exportPath}`);
+        console.log(`   View with: evaluclaude ui`);
+      }
+
       if (tracer) {
         const trace = tracer.finalize();
         const tracePath = await saveTrace(trace);

src/cli/commands/ui.ts

@@ -1,7 +1,7 @@
 import { Command } from 'commander';
-import { spawn, type ChildProcess } from 'child_process';
+import { spawn } from 'child_process';
 import { existsSync, readFileSync, writeFileSync, mkdirSync } from 'fs';
-import { join, dirname } from 'path';
+import { join, dirname, resolve as resolvePath } from 'path';
 import type { EvalSpec } from '../../analyzer/types.js';
 import { generatePromptfooConfig, generateTestProvider } from '../../promptfoo/index.js';
 
@@ -21,6 +21,7 @@ export const uiCommand = new Command('ui')
       const configPath = join(EVALUCLAUDE_DIR, CONFIG_FILE);
       const providerPath = join(EVALUCLAUDE_DIR, PROVIDERS_DIR, 'test-runner.py');
 
+      // If spec provided with --generate, create/update Promptfoo config
       if (options.spec && options.generate) {
         console.log('\n📄 Generating Promptfoo configuration...');
         
@@ -36,6 +37,7 @@ export const uiCommand = new Command('ui')
           outputPath: configPath,
           framework: detectFramework(spec),
           includeTraceLinks: true,
+          providerPath: providerPath,
         });
 
         await generateTestProvider(providerPath);
@@ -44,20 +46,31 @@ export const uiCommand = new Command('ui')
         console.log(`   Provider: ${providerPath}`);
       }
 
+      // Check for existing config, create default if missing
       if (!existsSync(configPath)) {
         console.log('\n⚠️  No Promptfoo config found.');
-        console.log('   Run with --spec <file> --generate to create one.\n');
-        console.log('   Or create one manually:');
-        console.log(`   ${configPath}\n`);
+        console.log('   Creating default configuration...\n');
         
         await createDefaultConfig(configPath, providerPath);
-        console.log(`   Created default config at ${configPath}`);
+        console.log(`   Created: ${configPath}`);
+      }
+
+      // Check for results to display
+      const resultsDir = join(EVALUCLAUDE_DIR, 'results');
+      const latestResults = join(resultsDir, 'latest.json');
+      
+      if (!existsSync(latestResults)) {
+        console.log('\n⚠️  No evaluation results found.');
+        console.log('   Run `evaluclaude run --export-promptfoo` first to generate results.\n');
+        console.log('   Or run the full pipeline:');
+        console.log('   evaluclaude pipeline <path> --promptfoo\n');
       }
 
       console.log(`\n🚀 Starting Promptfoo UI on port ${port}...`);
-      console.log(`   Config: ${configPath}\n`);
+      console.log(`   Results: ${latestResults}\n`);
 
-      await launchPromptfooUI(port, configPath, options.open);
+      // Use promptfoo view with the results file
+      await launchPromptfooView(port, latestResults, options.open);
     } catch (error) {
       console.error('Error launching UI:', error instanceof Error ? error.message : error);
       process.exit(1);
@@ -71,12 +84,21 @@ export const evalCommand = new Command('eval')
   .option('-o, --output <output>', 'Output path for results', '.evaluclaude/results')
   .option('--view', 'Launch UI after evaluation', false)
   .option('-p, --port <port>', 'Port for UI', '3000')
+  .option('--no-cache', 'Disable Promptfoo caching', false)
   .action(async (options) => {
     try {
       const configPath = options.config || join(EVALUCLAUDE_DIR, CONFIG_FILE);
+      const providerPath = join(EVALUCLAUDE_DIR, PROVIDERS_DIR, 'test-runner.py');
 
+      // Generate config from spec if provided
       if (options.spec) {
         console.log('\n📄 Generating Promptfoo configuration from spec...');
+        
+        if (!existsSync(options.spec)) {
+          console.error(`Error: Spec file not found: ${options.spec}`);
+          process.exit(1);
+        }
+
         const spec: EvalSpec = JSON.parse(readFileSync(options.spec, 'utf-8'));
         
         await generatePromptfooConfig(spec, {
@@ -84,30 +106,57 @@ export const evalCommand = new Command('eval')
           outputPath: configPath,
           framework: detectFramework(spec),
           includeTraceLinks: true,
+          providerPath: providerPath,
         });
 
-        const providerPath = join(EVALUCLAUDE_DIR, PROVIDERS_DIR, 'test-runner.py');
         await generateTestProvider(providerPath);
+        
+        console.log(`   Config: ${configPath}`);
+        console.log(`   Provider: ${providerPath}`);
+        console.log(`   Scenarios: ${spec.scenarios.length}`);
       }
 
       if (!existsSync(configPath)) {
-        console.error(`Error: Config not found: ${configPath}`);
-        console.log('Run with --spec <file> to generate from EvalSpec.');
+        console.error(`\nError: Config not found: ${configPath}`);
+        console.log('Run with --spec <file> to generate from EvalSpec, or create config manually.');
         process.exit(1);
       }
 
-      console.log('\n🧪 Running Promptfoo evaluations...\n');
+      // Ensure output directory exists
+      mkdirSync(options.output, { recursive: true });
+
+      console.log('\n🧪 Running Promptfoo evaluations...');
+      console.log(`   Config: ${configPath}`);
+      console.log(`   Output: ${options.output}\n`);
 
       const outputFile = join(options.output, `eval-${Date.now()}.json`);
-      mkdirSync(dirname(outputFile), { recursive: true });
 
-      await runPromptfooEval(configPath, outputFile);
+      const exitCode = await runPromptfooEval(configPath, outputFile, !options.cache);
 
-      console.log(`\n📁 Results saved: ${outputFile}`);
+      if (exitCode === 0) {
+        console.log(`\n✅ Evaluation complete!`);
+        console.log(`📁 Results: ${outputFile}`);
+      } else {
+        console.log(`\n⚠️  Evaluation finished with exit code ${exitCode}`);
+        console.log(`📁 Results: ${outputFile}`);
+      }
+
+      // List traces generated during evaluation
+      const tracesDir = join(EVALUCLAUDE_DIR, 'traces');
+      if (existsSync(tracesDir)) {
+        const { readdirSync } = await import('fs');
+        const traces = readdirSync(tracesDir).filter(f => f.endsWith('.json'));
+        if (traces.length > 0) {
+          console.log(`\n📊 Traces generated: ${traces.length}`);
+          console.log(`   View with: evaluclaude view --last`);
+        }
+      }
 
       if (options.view) {
         console.log(`\n🚀 Launching UI on port ${options.port}...`);
         await launchPromptfooUI(parseInt(options.port, 10), configPath, true);
+      } else {
+        console.log(`\n   View results: evaluclaude ui`);
       }
     } catch (error) {
       console.error('Error running eval:', error instanceof Error ? error.message : error);
@@ -115,6 +164,64 @@ export const evalCommand = new Command('eval')
     }
   });
 
+/**
+ * Launch Promptfoo view to display pre-computed results.
+ */
+async function launchPromptfooView(
+  port: number,
+  resultsFile: string,
+  openBrowser: boolean
+): Promise<void> {
+  return new Promise((resolve, reject) => {
+    // Use 'promptfoo view' which opens the web UI showing results from the output directory
+    const resultsDir = dirname(resolvePath(resultsFile));
+    const args = ['promptfoo', 'view', '--port', String(port)];
+    
+    if (openBrowser) {
+      args.push('-y');
+    } else {
+      args.push('-n');
+    }
+
+    // Pass the directory containing results
+    args.push(resultsDir);
+
+    console.log(`   Running: npx ${args.join(' ')}\n`);
+
+    const child = spawn('npx', args, {
+      stdio: 'inherit',
+      env: { ...process.env },
+    });
+
+    child.on('error', (error) => {
+      if ((error as NodeJS.ErrnoException).code === 'ENOENT') {
+        console.error('\n❌ Promptfoo not found.');
+        console.error('   Install with: npm install -g promptfoo');
+        console.error('   Or run: npx promptfoo --version\n');
+      } else {
+        reject(error);
+      }
+    });
+
+    child.on('close', (code) => {
+      if (code === 0) {
+        resolve();
+      } else {
+        reject(new Error(`Promptfoo exited with code ${code}`));
+      }
+    });
+
+    // Handle Ctrl+C gracefully
+    process.on('SIGINT', () => {
+      child.kill('SIGINT');
+      process.exit(0);
+    });
+  });
+}
+
+/**
+ * Launch Promptfoo with a config file (for running evals).
+ */
 async function launchPromptfooUI(
   port: number, 
   configPath: string, 
@@ -129,7 +236,8 @@ async function launchPromptfooUI(
       args.push('-n');
     }
 
-    const configDir = dirname(configPath);
+    // Pass the directory containing the config
+    const configDir = dirname(resolvePath(configPath));
     args.push(configDir);
 
     console.log(`   Running: npx ${args.join(' ')}\n`);
@@ -141,7 +249,9 @@ async function launchPromptfooUI(
 
     child.on('error', (error) => {
       if ((error as NodeJS.ErrnoException).code === 'ENOENT') {
-        console.error('\n❌ Promptfoo not found. Install with: npm install -g promptfoo');
+        console.error('\n❌ Promptfoo not found.');
+        console.error('   Install with: npm install -g promptfoo');
+        console.error('   Or run: npx promptfoo --version\n');
       } else {
         reject(error);
       }
@@ -155,6 +265,7 @@ async function launchPromptfooUI(
       }
     });
 
+    // Handle Ctrl+C gracefully
     process.on('SIGINT', () => {
       child.kill('SIGINT');
       process.exit(0);
@@ -162,16 +273,23 @@ async function launchPromptfooUI(
   });
 }
 
-async function runPromptfooEval(configPath: string, outputFile: string): Promise<void> {
+async function runPromptfooEval(
+  configPath: string, 
+  outputFile: string,
+  noCache: boolean
+): Promise<number> {
   return new Promise((resolve, reject) => {
     const args = [
       'promptfoo', 
       'eval', 
       '-c', configPath,
       '-o', outputFile,
-      '--no-cache',
     ];
 
+    if (noCache) {
+      args.push('--no-cache');
+    }
+
     console.log(`   Running: npx ${args.join(' ')}\n`);
 
     const child = spawn('npx', args, {
@@ -179,14 +297,18 @@ async function runPromptfooEval(configPath: string, outputFile: string): Promise
       env: { ...process.env },
     });
 
-    child.on('error', reject);
+    child.on('error', (error) => {
+      if ((error as NodeJS.ErrnoException).code === 'ENOENT') {
+        console.error('\n❌ Promptfoo not found.');
+        console.error('   Install with: npm install -g promptfoo\n');
+        reject(error);
+      } else {
+        reject(error);
+      }
+    });
 
     child.on('close', (code) => {
-      if (code === 0) {
-        resolve();
-      } else {
-        reject(new Error(`Promptfoo eval exited with code ${code}`));
-      }
+      resolve(code ?? 1);
     });
   });
 }
@@ -194,6 +316,14 @@ async function runPromptfooEval(configPath: string, outputFile: string): Promise
 async function createDefaultConfig(configPath: string, providerPath: string): Promise<void> {
   const defaultConfig = `# Evaluclaude Promptfoo Configuration
 # Generated by evaluclaude
+# 
+# To populate this config from an EvalSpec:
+#   evaluclaude eval --spec <evalspec.json>
+#
+# Or run the full pipeline:
+#   evaluclaude analyze <path> -o spec.json
+#   evaluclaude render spec.json -o tests/generated
+#   evaluclaude eval --spec spec.json
 
 description: "Evaluclaude functional test evaluations"
 
@@ -204,12 +334,13 @@ providers:
       test_dir: ./tests/generated
       framework: pytest
       timeout: 300
+      sandbox: true
 
 prompts:
   - "{{scenario_id}}"
 
 tests:
-  - description: "Example test"
+  - description: "Example test - replace with real scenarios"
     vars:
       scenario_id: "test_example"
     assert:
@@ -219,12 +350,19 @@ tests:
           result = json.loads(output)
           result.get('passed', 0) > 0
 
+# Default test configuration
+defaultTest:
+  metadata:
+    evaluclaude: true
+    tracesDir: .evaluclaude/traces
+
 outputPath: .evaluclaude/results/promptfoo-results.json
 `;
 
   mkdirSync(dirname(configPath), { recursive: true });
   writeFileSync(configPath, defaultConfig);
 
+  // Also generate the provider
   await generateTestProvider(providerPath);
 }
 
@@ -232,5 +370,8 @@ function detectFramework(spec: EvalSpec): 'pytest' | 'vitest' | 'jest' {
   if (spec.repo.languages.includes('python')) {
     return 'pytest';
   }
+  if (spec.repo.languages.includes('typescript') || spec.repo.languages.includes('javascript')) {
+    return 'vitest';
+  }
   return 'vitest';
 }

src/cli/index.ts

@@ -8,6 +8,7 @@ import { gradeCommand, listRubricsCommand, calibrateCommand } from './commands/g
 import { runCommand } from './commands/run.js';
 import { viewCommand, tracesCommand } from './commands/view.js';
 import { uiCommand, evalCommand } from './commands/ui.js';
+import { pipelineCommand } from './commands/pipeline.js';
 
 const program = new Command();
 
@@ -16,15 +17,25 @@ program
   .description('Zero-to-evals in one command. Claude analyzes codebases and generates functional tests.')
   .version('0.1.0');
 
+// Core pipeline command - the "zero to evals" experience
+program.addCommand(pipelineCommand);
+
+// Individual step commands
 program.addCommand(introCommand);
 program.addCommand(analyzeCommand);
 program.addCommand(renderCommand);
+program.addCommand(runCommand);
+
+// Grading commands
 program.addCommand(gradeCommand);
 program.addCommand(listRubricsCommand);
 program.addCommand(calibrateCommand);
-program.addCommand(runCommand);
+
+// Observability commands
 program.addCommand(viewCommand);
 program.addCommand(tracesCommand);
+
+// Promptfoo integration commands
 program.addCommand(uiCommand);
 program.addCommand(evalCommand);
 

src/observability/trace-viewer.ts

@@ -64,7 +64,7 @@ export function formatTrace(trace: EvalTrace, options: Partial<ViewOptions> = {}
   lines.push('─'.repeat(40));
   lines.push(`   ✅ Passed:         ${trace.execution.testsPassed}`);
   lines.push(`   ❌ Failed:         ${trace.execution.testsFailed}`);
-  lines.push(`   ⏭️  Skipped:        ${trace.execution.testsSkipped}`);
+  lines.push(`   ⏭️  Skipped:        ${trace.execution.testsSkipped ?? 0}`);
   lines.push('');
 
   if (opts.showQuestions && trace.analysis.questionsAsked.length > 0) {

src/promptfoo/config-generator.ts

@@ -1,5 +1,5 @@
 import { writeFile, mkdir } from 'fs/promises';
-import { dirname, join } from 'path';
+import { dirname, join, resolve } from 'path';
 import * as yaml from 'js-yaml';
 import type { EvalSpec, EvalScenario } from '../analyzer/types.js';
 import type { PromptfooConfig, PromptfooTest, PromptfooAssertion } from './types.js';
@@ -9,6 +9,7 @@ export interface ConfigOptions {
   outputPath: string;
   framework: 'pytest' | 'vitest' | 'jest';
   includeTraceLinks: boolean;
+  providerPath?: string;
 }
 
 export async function generatePromptfooConfig(
@@ -29,17 +30,24 @@ export async function generatePromptfooConfig(
 
 function buildConfig(spec: EvalSpec, options: ConfigOptions): PromptfooConfig {
   const tests = spec.scenarios.map(scenario => buildTest(scenario, options));
+  
+  // Provider path should be relative to the config file location
+  // Since config is at .evaluclaude/promptfooconfig.yaml, the provider is at ./providers/test-runner.py
+  const providerRelativePath = options.providerPath 
+    ? options.providerPath.replace('.evaluclaude/', './').replace(/^\.evaluclaude\//, './')
+    : './providers/test-runner.py';
 
   return {
     description: `Evaluclaude functional tests for ${spec.repo.name}`,
     providers: [
       {
-        id: `file://providers/test-runner.py`,
+        id: `file://${providerRelativePath}`,
         label: 'functional-tests',
         config: {
-          test_dir: options.testDir,
+          test_dir: resolve(options.testDir),
           framework: options.framework,
           timeout: 300,
+          sandbox: true,
         },
       },
     ],
@@ -48,11 +56,12 @@ function buildConfig(spec: EvalSpec, options: ConfigOptions): PromptfooConfig {
     defaultTest: options.includeTraceLinks
       ? {
           metadata: {
-            traceFile: '.evaluclaude/traces/{{evalId}}.json',
+            evaluclaude: true,
+            tracesDir: './traces',
           },
         }
       : undefined,
-    outputPath: '.evaluclaude/results/promptfoo-results.json',
+    outputPath: './results/promptfoo-results.json',
   };
 }
 
@@ -147,92 +156,51 @@ function buildAssertion(assertion: any): PromptfooAssertion {
 
 export async function generateTestProvider(outputPath: string): Promise<void> {
   const providerCode = `#!/usr/bin/env python3
-"""Promptfoo provider that executes tests and returns structured results."""
+"""
+Promptfoo provider that executes tests and returns structured results.
+
+This provider integrates with evaluclaude-harness test runners to execute
+functional tests in a sandboxed environment and return results compatible
+with Promptfoo's assertion system.
+"""
 
 import subprocess
 import json
 import sys
 import os
+import tempfile
+import uuid
+from pathlib import Path
 
-def get_provider_response(prompt: str, options: dict, context: dict) -> dict:
+
+def call_api(prompt: str, options: dict, context: dict) -> dict:
     """Runs tests and returns structured results."""
     
-    test_dir = options.get('config', {}).get('test_dir', './tests')
-    framework = options.get('config', {}).get('framework', 'pytest')
-    timeout = options.get('config', {}).get('timeout', 300)
+    config = options.get('config', {})
+    test_dir = config.get('test_dir', './tests/generated')
+    framework = config.get('framework', 'pytest')
+    timeout = config.get('timeout', 300)
+    sandbox = config.get('sandbox', True)
     
     scenario_id = prompt.strip()
+    eval_id = f"eval-{uuid.uuid4().hex[:8]}"
+    
+    # Ensure traces directory exists
+    traces_dir = Path('.evaluclaude/traces')
+    traces_dir.mkdir(parents=True, exist_ok=True)
     
     try:
         if framework == 'pytest':
-            result = subprocess.run(
-                [
-                    'python', '-m', 'pytest',
-                    '--json-report',
-                    '--json-report-file=/tmp/pytest_results.json',
-                    '-k', scenario_id,
-                    test_dir
-                ],
-                capture_output=True,
-                text=True,
-                timeout=timeout
-            )
-            
-            try:
-                with open('/tmp/pytest_results.json') as f:
-                    report = json.load(f)
-                    
-                output = {
-                    'passed': report.get('summary', {}).get('passed', 0),
-                    'failed': report.get('summary', {}).get('failed', 0),
-                    'skipped': report.get('summary', {}).get('skipped', 0),
-                    'tests': report.get('tests', []),
-                    'stdout': result.stdout,
-                    'stderr': result.stderr,
-                    'exit_code': result.returncode,
-                }
-            except FileNotFoundError:
-                output = {
-                    'passed': 0,
-                    'failed': 1,
-                    'error': 'Failed to generate pytest report',
-                    'stdout': result.stdout,
-                    'stderr': result.stderr,
-                }
-                
+            output = run_pytest(test_dir, scenario_id, timeout, eval_id)
         elif framework in ('vitest', 'jest'):
-            cmd = ['npx', framework, 'run', '--reporter=json']
-            if scenario_id:
-                cmd.extend(['--testNamePattern', scenario_id])
-            cmd.append(test_dir)
-            
-            result = subprocess.run(
-                cmd,
-                capture_output=True,
-                text=True,
-                timeout=timeout
-            )
-            
-            try:
-                report = json.loads(result.stdout)
-                output = {
-                    'passed': report.get('numPassedTests', 0),
-                    'failed': report.get('numFailedTests', 0),
-                    'skipped': report.get('numSkippedTests', 0),
-                    'tests': report.get('testResults', []),
-                    'exit_code': result.returncode,
-                }
-            except json.JSONDecodeError:
-                output = {
-                    'passed': 0,
-                    'failed': 1,
-                    'error': 'Failed to parse test output',
-                    'stdout': result.stdout,
-                    'stderr': result.stderr,
-                }
+            output = run_js_tests(test_dir, scenario_id, timeout, framework, eval_id)
         else:
-            output = {'error': f'Unknown framework: {framework}'}
+            output = {'error': f'Unknown framework: {framework}', 'passed': 0, 'failed': 1}
             
+        # Add trace reference
+        output['eval_id'] = eval_id
+        output['trace_file'] = str(traces_dir / f"{eval_id}.json")
+        
         return {
             'output': json.dumps(output),
             'error': None,
@@ -240,32 +208,187 @@ def get_provider_response(prompt: str, options: dict, context: dict) -> dict:
         
     except subprocess.TimeoutExpired:
         return {
-            'output': json.dumps({'error': 'Test execution timed out', 'passed': 0, 'failed': 1}),
+            'output': json.dumps({
+                'error': 'Test execution timed out',
+                'passed': 0,
+                'failed': 1,
+                'eval_id': eval_id,
+            }),
             'error': None,
         }
     except Exception as e:
         return {
-            'output': None,
+            'output': json.dumps({
+                'error': str(e),
+                'passed': 0,
+                'failed': 1,
+                'eval_id': eval_id,
+            }),
             'error': str(e),
         }
 
+
+def run_pytest(test_dir: str, scenario_id: str, timeout: int, eval_id: str) -> dict:
+    """Run pytest and return structured results."""
+    with tempfile.NamedTemporaryFile(suffix='.json', delete=False) as f:
+        report_file = f.name
+    
+    cmd = [
+        sys.executable, '-m', 'pytest',
+        '--json-report',
+        f'--json-report-file={report_file}',
+        '-v',
+        '--tb=short',
+    ]
+    
+    if scenario_id:
+        cmd.extend(['-k', scenario_id])
+    
+    cmd.append(test_dir)
+    
+    result = subprocess.run(
+        cmd,
+        capture_output=True,
+        text=True,
+        timeout=timeout,
+        cwd=os.getcwd(),
+    )
+    
+    try:
+        with open(report_file) as f:
+            report = json.load(f)
+            
+        summary = report.get('summary', {})
+        tests = report.get('tests', [])
+        
+        output = {
+            'passed': summary.get('passed', 0),
+            'failed': summary.get('failed', 0),
+            'skipped': summary.get('skipped', 0),
+            'total': summary.get('total', 0),
+            'duration': report.get('duration', 0) * 1000,  # Convert to ms
+            'tests': [
+                {
+                    'id': extract_scenario_id(t.get('nodeid', '')),
+                    'name': t.get('nodeid', ''),
+                    'status': t.get('outcome', 'unknown'),
+                    'duration': (t.get('call', {}).get('duration', 0) or 0) * 1000,
+                    'error': t.get('call', {}).get('crash', {}).get('message') if t.get('call', {}).get('crash') else None,
+                }
+                for t in tests
+            ],
+            'exit_code': result.returncode,
+        }
+    except (FileNotFoundError, json.JSONDecodeError) as e:
+        output = {
+            'passed': 0,
+            'failed': 1,
+            'error': f'Failed to parse pytest report: {e}',
+            'stdout': result.stdout[-2000:] if result.stdout else '',
+            'stderr': result.stderr[-2000:] if result.stderr else '',
+        }
+    finally:
+        try:
+            os.unlink(report_file)
+        except OSError:
+            pass
+    
+    return output
+
+
+def run_js_tests(test_dir: str, scenario_id: str, timeout: int, framework: str, eval_id: str) -> dict:
+    """Run vitest/jest and return structured results."""
+    with tempfile.NamedTemporaryFile(suffix='.json', delete=False) as f:
+        report_file = f.name
+    
+    cmd = ['npx', framework, 'run', '--reporter=json', f'--outputFile={report_file}']
+    
+    if scenario_id:
+        cmd.extend(['--testNamePattern', scenario_id])
+    
+    cmd.append(test_dir)
+    
+    result = subprocess.run(
+        cmd,
+        capture_output=True,
+        text=True,
+        timeout=timeout,
+        cwd=os.getcwd(),
+    )
+    
+    try:
+        with open(report_file) as f:
+            report = json.load(f)
+        
+        output = {
+            'passed': report.get('numPassedTests', 0),
+            'failed': report.get('numFailedTests', 0),
+            'skipped': report.get('numSkippedTests', 0),
+            'total': report.get('numTotalTests', 0),
+            'tests': [],
+            'exit_code': result.returncode,
+        }
+        
+        for test_file in report.get('testResults', []):
+            for assertion in test_file.get('assertionResults', []):
+                output['tests'].append({
+                    'id': extract_scenario_id(assertion.get('fullName', '')),
+                    'name': assertion.get('fullName', ''),
+                    'status': assertion.get('status', 'unknown'),
+                    'duration': assertion.get('duration', 0),
+                    'error': assertion.get('failureMessages', [None])[0] if assertion.get('failureMessages') else None,
+                })
+                
+    except (FileNotFoundError, json.JSONDecodeError) as e:
+        output = {
+            'passed': 0,
+            'failed': 1,
+            'error': f'Failed to parse {framework} report: {e}',
+            'stdout': result.stdout[-2000:] if result.stdout else '',
+            'stderr': result.stderr[-2000:] if result.stderr else '',
+        }
+    finally:
+        try:
+            os.unlink(report_file)
+        except OSError:
+            pass
+    
+    return output
+
+
+def extract_scenario_id(nodeid: str) -> str:
+    """Extract scenario ID from test name."""
+    import re
+    match = re.search(r'test[_\\s]([a-zA-Z0-9_-]+)', nodeid, re.IGNORECASE)
+    return match.group(1) if match else nodeid.replace(' ', '_')
+
+
+def get_provider_response(prompt: str, options: dict, context: dict) -> dict:
+    """Alias for call_api for backwards compatibility."""
+    return call_api(prompt, options, context)
+
+
 if __name__ == '__main__':
-    # For testing the provider directly
     import argparse
-    parser = argparse.ArgumentParser()
-    parser.add_argument('--scenario', default='')
-    parser.add_argument('--test-dir', default='./tests')
-    parser.add_argument('--framework', default='pytest')
+    parser = argparse.ArgumentParser(description='Run tests for Promptfoo')
+    parser.add_argument('--scenario', default='', help='Scenario ID to filter')
+    parser.add_argument('--test-dir', default='./tests/generated', help='Test directory')
+    parser.add_argument('--framework', default='pytest', help='Test framework')
+    parser.add_argument('--timeout', type=int, default=300, help='Timeout in seconds')
     args = parser.parse_args()
     
-    result = get_provider_response(
+    result = call_api(
         args.scenario,
-        {'config': {'test_dir': args.test_dir, 'framework': args.framework}},
+        {'config': {
+            'test_dir': args.test_dir,
+            'framework': args.framework,
+            'timeout': args.timeout,
+        }},
         {}
     )
-    print(json.dumps(result, indent=2))
+    print(json.dumps(json.loads(result['output']), indent=2) if result['output'] else result['error'])
 `;
 
   await mkdir(dirname(outputPath), { recursive: true });
-  await writeFile(outputPath, providerCode);
+  await writeFile(outputPath, providerCode, { mode: 0o755 });
 }

src/promptfoo/index.ts

@@ -1,2 +1,13 @@
 export * from './types.js';
-export { generatePromptfooConfig, generateTestProvider } from './config-generator.js';
+export { generatePromptfooConfig, generateTestProvider, type ConfigOptions } from './config-generator.js';
+export { 
+  runTestsForPromptfoo, 
+  savePromptfooResults,
+  type RunTestsForPromptfooOptions,
+  type PromptfooProviderResult,
+} from './runner-bridge.js';
+export {
+  exportToPromptfooFormat,
+  generateViewOnlyConfig,
+  type ExportOptions,
+} from './results-exporter.js';

src/promptfoo/results-exporter.ts

@@ -0,0 +1,127 @@
+/**
+ * Export test execution results to Promptfoo format for viewing in the UI.
+ * 
+ * Instead of using Promptfoo to run tests (which requires a provider that
+ * responds quickly), we run tests ourselves and export results to Promptfoo's
+ * result format. This allows us to use Promptfoo's excellent visualization UI.
+ */
+
+import { writeFile, mkdir } from 'fs/promises';
+import { join } from 'path';
+import type { ExecutionResult } from '../runners/types.js';
+import type { EvalSpec } from '../analyzer/types.js';
+import type { PromptfooResult, PromptfooTestResult } from './types.js';
+
+export interface ExportOptions {
+  outputDir: string;
+  evalId?: string;
+  includeSpec?: boolean;
+}
+
+/**
+ * Export ExecutionResult to Promptfoo result format.
+ */
+export async function exportToPromptfooFormat(
+  result: ExecutionResult,
+  spec: EvalSpec | undefined,
+  options: ExportOptions
+): Promise<string> {
+  const { outputDir, evalId = `eval-${Date.now()}` } = options;
+
+  const promptfooResult = buildPromptfooResult(result, spec, evalId);
+
+  await mkdir(outputDir, { recursive: true });
+  const outputPath = join(outputDir, `${evalId}.json`);
+  await writeFile(outputPath, JSON.stringify(promptfooResult, null, 2));
+
+  // Also write the latest.json symlink equivalent
+  const latestPath = join(outputDir, 'latest.json');
+  await writeFile(latestPath, JSON.stringify(promptfooResult, null, 2));
+
+  return outputPath;
+}
+
+function buildPromptfooResult(
+  result: ExecutionResult,
+  spec: EvalSpec | undefined,
+  evalId: string
+): PromptfooResult {
+  const testResults: PromptfooTestResult[] = result.tests.map(test => {
+    // Try to find matching scenario from spec
+    const scenario = spec?.scenarios.find(s => 
+      s.id === test.id || test.name.includes(s.id)
+    );
+
+    return {
+      prompt: {
+        raw: scenario?.id || test.id,
+        label: scenario?.name || test.name,
+      },
+      vars: {
+        scenario_id: scenario?.id || test.id,
+        target_module: scenario?.target.module || '',
+        target_function: scenario?.target.function || '',
+        description: scenario?.description || test.name,
+      },
+      response: {
+        output: test.status === 'passed' 
+          ? 'Test passed successfully'
+          : test.error?.message || 'Test failed',
+      },
+      gradingResult: {
+        pass: test.status === 'passed',
+        score: test.status === 'passed' ? 1 : 0,
+        reason: test.status === 'passed'
+          ? 'All assertions passed'
+          : test.error?.message || 'Test failed',
+        componentResults: test.assertions.details.map(a => ({
+          pass: a.passed,
+          score: a.passed ? 1 : 0,
+          reason: a.description,
+          assertion: {
+            type: 'custom',
+            value: a.description,
+          },
+        })),
+      },
+      success: test.status === 'passed',
+      error: test.error?.message,
+    };
+  });
+
+  return {
+    version: 1,
+    timestamp: new Date().toISOString(),
+    results: testResults,
+    stats: {
+      successes: result.summary.passed,
+      failures: result.summary.failed,
+      tokenUsage: {
+        total: 0,
+        prompt: 0,
+        completion: 0,
+      },
+    },
+  };
+}
+
+/**
+ * Generate a minimal Promptfoo config that just views results (no provider).
+ */
+export function generateViewOnlyConfig(spec: EvalSpec): string {
+  return `# Evaluclaude Results Config
+# This config is for viewing results only - tests are run via evaluclaude run
+
+description: "Test results for ${spec.repo.name}"
+
+# No providers needed - we pre-run tests and import results
+providers: []
+
+prompts: []
+
+tests: []
+
+# Results are stored here by evaluclaude run --export-promptfoo
+outputPath: .evaluclaude/results/latest.json
+`;
+}

src/promptfoo/runner-bridge.ts

@@ -0,0 +1,194 @@
+/**
+ * Bridge between our test runners and Promptfoo's provider interface.
+ * 
+ * This module provides a unified way to run tests that works both:
+ * 1. Standalone via our `run` command
+ * 2. As a Promptfoo provider via the generated test-runner.py
+ * 
+ * Results are stored in a format compatible with Promptfoo's expectations.
+ */
+
+import { writeFile, mkdir } from 'fs/promises';
+import { join, dirname } from 'path';
+import { runTests, type ExecutionResult, type ExecutionOptions, DEFAULT_SANDBOX_CONFIG } from '../runners/index.js';
+import { createTracer, saveTrace, type EvalTrace } from '../observability/index.js';
+
+export interface PromptfooProviderResult {
+  output: string;
+  error: string | null;
+  tokenUsage?: {
+    total: number;
+    prompt: number;
+    completion: number;
+  };
+}
+
+export interface RunTestsForPromptfooOptions {
+  scenarioId: string;
+  testDir: string;
+  framework: 'pytest' | 'vitest' | 'jest';
+  timeout?: number;
+  sandbox?: boolean;
+  evalId?: string;
+  recordTrace?: boolean;
+}
+
+/**
+ * Run tests for a specific scenario and format results for Promptfoo.
+ */
+export async function runTestsForPromptfoo(
+  options: RunTestsForPromptfooOptions
+): Promise<PromptfooProviderResult> {
+  const {
+    scenarioId,
+    testDir,
+    framework,
+    timeout = 300000,
+    sandbox = true,
+    evalId = `eval-${Date.now()}`,
+    recordTrace = true,
+  } = options;
+
+  const tracer = recordTrace ? createTracer(evalId) : null;
+
+  try {
+    const execOptions: ExecutionOptions = {
+      framework,
+      sandbox,
+      timeout,
+      parallel: false,
+      filter: scenarioId ? [scenarioId] : undefined,
+      cwd: process.cwd(),
+    };
+
+    tracer?.recordIntrospection({
+      filesAnalyzed: [testDir],
+      duration: 0,
+    });
+
+    const result = await runTests(
+      testDir,
+      execOptions,
+      sandbox ? DEFAULT_SANDBOX_CONFIG : undefined
+    );
+
+    // Record execution results in trace
+    if (tracer) {
+      tracer.recordExecution({
+        testsPassed: result.summary.passed,
+        testsFailed: result.summary.failed,
+        testsSkipped: result.summary.skipped,
+      });
+
+      for (const test of result.tests) {
+        if (test.status === 'failed' || test.status === 'error') {
+          tracer.recordTestFailure({
+            scenarioId: test.id,
+            testName: test.name,
+            error: test.error?.message || 'Unknown error',
+            stack: test.error?.stack,
+          });
+        }
+      }
+    }
+
+    // Build Promptfoo-compatible output
+    const promptfooOutput = buildPromptfooOutput(result, scenarioId);
+
+    // Save trace if enabled
+    if (tracer) {
+      const trace = tracer.finalize();
+      await saveTrace(trace);
+    }
+
+    return {
+      output: JSON.stringify(promptfooOutput),
+      error: null,
+    };
+  } catch (error) {
+    if (tracer) {
+      tracer.recordError(error instanceof Error ? error : new Error(String(error)));
+      const trace = tracer.finalize();
+      await saveTrace(trace);
+    }
+
+    return {
+      output: JSON.stringify({
+        passed: 0,
+        failed: 1,
+        error: error instanceof Error ? error.message : String(error),
+      }),
+      error: error instanceof Error ? error.message : String(error),
+    };
+  }
+}
+
+/**
+ * Build Promptfoo-compatible output from ExecutionResult.
+ */
+function buildPromptfooOutput(
+  result: ExecutionResult,
+  scenarioId?: string
+): Record<string, unknown> {
+  const matchingTests = scenarioId
+    ? result.tests.filter(t => t.id === scenarioId || t.name.includes(scenarioId))
+    : result.tests;
+
+  return {
+    passed: matchingTests.filter(t => t.status === 'passed').length,
+    failed: matchingTests.filter(t => t.status === 'failed' || t.status === 'error').length,
+    skipped: matchingTests.filter(t => t.status === 'skipped').length,
+    total: matchingTests.length,
+    tests: matchingTests.map(t => ({
+      id: t.id,
+      name: t.name,
+      status: t.status,
+      duration: t.duration,
+      error: t.error?.message,
+    })),
+    summary: {
+      ...result.summary,
+      matchedScenario: scenarioId,
+    },
+    errors: result.errors,
+  };
+}
+
+/**
+ * Generate a Promptfoo-compatible results file from our execution results.
+ */
+export async function savePromptfooResults(
+  result: ExecutionResult,
+  evalId: string,
+  outputDir: string = '.evaluclaude/results'
+): Promise<string> {
+  const promptfooResult = {
+    version: 1,
+    timestamp: new Date().toISOString(),
+    evalId,
+    results: result.tests.map(t => ({
+      prompt: { raw: t.id, label: t.name },
+      vars: { scenario_id: t.id },
+      response: {
+        output: t.status === 'passed' ? 'PASS' : t.error?.message || 'FAIL',
+      },
+      gradingResult: {
+        pass: t.status === 'passed',
+        score: t.status === 'passed' ? 1 : 0,
+        reason: t.error?.message || (t.status === 'passed' ? 'Test passed' : 'Test failed'),
+      },
+      success: t.status === 'passed',
+      error: t.error?.message,
+    })),
+    stats: {
+      successes: result.summary.passed,
+      failures: result.summary.failed,
+    },
+  };
+
+  await mkdir(outputDir, { recursive: true });
+  const outputPath = join(outputDir, `promptfoo-${evalId}.json`);
+  await writeFile(outputPath, JSON.stringify(promptfooResult, null, 2));
+
+  return outputPath;
+}

src/runners/index.ts

@@ -73,8 +73,8 @@ export function formatResults(result: ExecutionResult): string {
   lines.push(`   Total:   ${result.summary.total}`);
   lines.push(`   ✅ Passed:  ${result.summary.passed}`);
   lines.push(`   ❌ Failed:  ${result.summary.failed}`);
-  lines.push(`   ⏭️  Skipped: ${result.summary.skipped}`);
-  lines.push(`   ⏱️  Duration: ${result.summary.duration}ms`);
+  lines.push(`   ⏭️  Skipped: ${result.summary.skipped ?? 0}`);
+  lines.push(`   ⏱️  Duration: ${result.summary.duration || 0}ms`);
   
   if (result.errors.length > 0) {
     lines.push('');