threadlines 0.2.25 → 0.3.0

package/README.md

@@ -4,17 +4,17 @@ Threadline CLI - AI-powered linter based on your natural language documentation.
 
 ## Why Threadline?
 
-Getting teams to follow consistent quality standards is **hard**. Really hard.
+Getting teams to consistently follow coding patterns and quality standards is **hard**. Really hard.
 
 - **Documentation** → Nobody reads it. Or it's outdated before you finish writing it.
 - **Linting** → Catches syntax errors, but misses nuanced stuff.
 - **AI Code Reviewers** → Powerful, but you can't trust them. Did they actually check what you care about? Can you customize them with your team's specific rules?
 
-**Threadline solves this** by running **separate, parallel, highly focused AI-powered reviews** - each focused on a single, specific concern. Your coding standards live in your repository as markdown files, version-controlled and always in sync with your codebase. Each threadline gets its own dedicated AI check, ensuring focused attention on what matters to your team.
+**Threadline solves this** by running **separate, parallel, highly focused AI-powered reviews** - each focused on a single, specific concern or pattern: the stuff that takes engineers months to internalise - and they keep forgetting. Your coding patterns live in your repository as 'Threadline' markdown files, version-controlled and always in sync with your codebase. Each threadline is its own AI agent, ensuring focused attention on what matters to your team.
 
 ### What Makes Threadline Different?
 
-- **Focused Reviews** - Instead of one AI trying to check everything, Threadline runs multiple specialized AI reviewers in parallel. Each threadline focuses on one thing and does it well.
+- **Focused Reviews** - Instead of one AI agent checking everything, Threadline runs multiple specialized AI reviewers in parallel. Each threadline focuses on one thing and does it well.
 
 - **Documentation That Lives With Your Code** - Your coding standards live in your repo, in a `/threadlines` folder. They're version-controlled, reviewable, and always in sync with your codebase.
 
@@ -24,18 +24,7 @@ Getting teams to follow consistent quality standards is **hard**. Really hard.
 
 ## Installation
 
-### Option 1: Global Installation (Recommended for Regular Use)
-
-```bash
-npm install -g threadlines
-```
-
-Then use directly:
-```bash
-threadlines check
-```
-
-### Option 2: Use with npx (No Installation)
+### Option 1: Use with npx (No Installation)
 
 ```bash
 npx threadlines check
@@ -82,8 +71,17 @@ Edit `threadlines/example.md` with your coding standards, then rename it to some
 ```bash
 npx threadlines check
 ```
+
 By default, analyzes your staged/unstaged git changes against all threadlines in the `/threadlines` directory.
 
+**Review Context Types:**
+- `local` - Staged/unstaged changes (default for local development)
+- `commit` - Specific commit (when using `--commit` flag)
+- `pr` - Pull Request/Merge Request (auto-detected in CI)
+- `file` - Single file (when using `--file` flag)
+- `folder` - Folder contents (when using `--folder` flag)
+- `files` - Multiple files (when using `--files` flag)
+
 **Common Use Cases:**
 
 **Check latest commit locally:**
@@ -96,11 +94,6 @@ threadlines check --commit HEAD
 threadlines check --commit abc123def
 ```
 
-**Check all commits in a branch:**
-```bash
-threadlines check --branch feature/new-feature
-```
-
 **Check entire file(s):**
 ```bash
 threadlines check --file src/api/users.ts
@@ -108,24 +101,35 @@ threadlines check --files src/api/users.ts src/api/posts.ts
 threadlines check --folder src/api
 ```
 
+**Debug mode (verbose output):**
+```bash
+threadlines check --debug
+```
+
 **Show all results (not just violations):**
 ```bash
 threadlines check --full
 ```
 
+**Enable debug logging:**
+```bash
+threadlines check --debug
+```
+
 **Options:**
-- `--api-url <url>` - Override the server URL (default: http://localhost:3000)
-- `--commit <ref>` - Review specific commit. Accepts commit SHA or git reference (e.g., `HEAD`, `HEAD~1`, `abc123`)
-- `--branch <name>` - Review all commits in branch vs base
-- `--file <path>` - Review entire file (all lines as additions)
-- `--folder <path>` - Review all files in folder recursively
-- `--files <paths...>` - Review multiple specified files
+- `--commit <ref>` - Review specific commit. Accepts commit SHA or git reference (e.g., `HEAD`, `HEAD~1`, `abc123`). Sets review context to `commit`.
+- `--file <path>` - Review entire file (all lines as additions). Sets review context to `file`.
+- `--folder <path>` - Review all files in folder recursively. Sets review context to `folder`.
+- `--files <paths...>` - Review multiple specified files. Sets review context to `files`.
 - `--full` - Show all results (compliant, attention, not_relevant). Default: only attention items
+- `--debug` - Enable debug logging (verbose output for troubleshooting)
+
+**Note:** Flags (`--commit`, `--file`, `--folder`, `--files`) are for local development only. In CI/CD environments, these flags are ignored and the CLI auto-detects the appropriate context.
 
 **Auto-detection in CI:**
-- CI with branch detected → reviews all commits in branch vs base
-- CI with commit SHA detected → reviews specific commit
-- Local development → reviews staged/unstaged changes
+- **Pull Request/Merge Request context** → Reviews all changes in the PR/MR (review context: `pr`)
+- **Push to any branch** → Reviews the commit being pushed (review context: `commit`)
+- **Local development** → Reviews staged/unstaged changes (review context: `local`)
 
 ## Configuration
 
@@ -135,11 +139,40 @@ threadlines check --full
 |----------|---------|----------|
 | `THREADLINE_API_KEY` | Authentication with Threadlines API | Yes |
 | `THREADLINE_ACCOUNT` | Your Threadlines account email | Yes |
-| `THREADLINE_API_URL` | Custom API endpoint (default: https://devthreadline.com) | No |
 
 Both required variables can be set in a `.env.local` file (recommended for local development) or as environment variables (required for CI/CD).
 
-The optional `THREADLINE_API_URL` allows you to point to a custom server if you want to intercept telemetry with your own endpoint. It can also be set with the `--api-url` flag.
+**Local Development:**
+Create a `.env.local` file in your project root:
+```bash
+THREADLINE_API_KEY=your-api-key-here
+THREADLINE_ACCOUNT=your-email@example.com
+```
+
+**CI/CD:**
+Set these as environment variables in your platform:
+- **GitHub Actions**: Settings → Secrets → Add variables
+- **GitLab CI**: Settings → CI/CD → Variables
+- **Bitbucket Pipelines**: Repository settings → Repository variables
+- **Vercel**: Settings → Environment Variables
+
+Get your credentials at: https://devthreadline.com/settings
+
+### Configuration File (`.threadlinerc`)
+
+You can customize the API endpoint and other settings by creating a `.threadlinerc` file in your project root:
+
+```json
+{
+  "mode": "online",
+  "api_url": "https://devthreadline.com",
+  "openai_model": "gpt-5.2",
+  "openai_service_tier": "Flex",
+  "diff_context_lines": 10
+}
+```
+
+The `api_url` field allows you to point to a custom server if needed. Default is `https://devthreadline.com`.
 
 ## Threadline Files
 
@@ -176,26 +209,38 @@ Your guidelines and standards here...
 
 - **`context_files`**: Array of file paths that provide context (always included, even if unchanged)
 
-### Example: SQL Queries with Schema Context
+### Example: Feature Flagging Standards
 
 ```markdown
 ---
-id: sql-queries
+id: feature-flags
 version: 1.0.0
 patterns:
-  - "**/queries/**"
-  - "**/*.sql"
+  - "**/features/**"
+  - "**/components/**"
+  - "**/*.tsx"
+  - "**/*.ts"
 context_files:
-  - "schema.sql"
+  - "config/feature-flags.ts"
 ---
 
-# SQL Query Standards
+# Feature Flag Standards
+
+All feature flag usage must:
+- Check flags using the centralized `isFeatureEnabled()` function from `config/feature-flags.ts`
+- Never hardcode feature flag names as strings (use constants from the config)
+- Include proper cleanup: remove feature flag checks when features are fully rolled out
+- Document rollout plan in PR description (target percentage, timeline)
+- Use feature flags for gradual rollouts, not as permanent configuration
 
-All SQL queries must:
-- Reference tables and columns that exist in schema.sql
-- Use parameterized queries (no string concatenation)
-- Include proper indexes for WHERE clauses
+**Violations:**
+- ❌ `if (process.env.NEW_FEATURE === 'true')` (hardcoded, not using registry)
+- ❌ `if (flags['new-feature'])` (string literal instead of constant)
+- ✅ `if (isFeatureEnabled(FeatureFlags.NEW_DASHBOARD))` (using centralized function)
 ```
 
-The `schema.sql` file will always be included as context, even if you're only changing query files.
+The `config/feature-flags.ts` file will always be included as context, ensuring the AI reviewer can verify that:
+- Feature flag names match the registry
+- The correct flag checking function is used
+- Flags are properly typed and documented
 

package/dist/api/client.js

@@ -19,5 +19,9 @@ class ReviewAPIClient {
         const response = await this.client.post('/api/threadline-check', request);
         return response.data;
     }
+    async syncResults(request) {
+        const response = await this.client.post('/api/threadline-check-results', request);
+        return response.data;
+    }
 }
 exports.ReviewAPIClient = ReviewAPIClient;

package/dist/commands/check.js

@@ -46,6 +46,7 @@ const ci_context_1 = require("../git/ci-context");
 const local_1 = require("../git/local");
 const config_file_1 = require("../utils/config-file");
 const logger_1 = require("../utils/logger");
+const expert_1 = require("../processors/expert");
 const fs = __importStar(require("fs"));
 const path = __importStar(require("path"));
 const chalk_1 = __importDefault(require("chalk"));
@@ -78,7 +79,7 @@ async function checkCommand(options) {
     const repoRoot = cwd; // Keep for backward compatibility with rest of function
     // Load configuration
     const config = await (0, config_file_1.loadConfig)(cwd);
-    console.log(chalk_1.default.blue(`🔍 Threadline CLI v${CLI_VERSION}: Checking code against your threadlines...\n`));
+    logger_1.logger.info(`🔍 Threadline CLI v${CLI_VERSION}: Checking code against your threadlines...\n`);
     // Get git root for consistent file paths across monorepo
     const git = (0, simple_git_1.default)(cwd);
     let gitRoot;
@@ -95,51 +96,36 @@ async function checkCommand(options) {
         logger_1.logger.error(`Failed to get git root: ${message}`);
         process.exit(1);
     }
-    // Pre-flight check: Validate ALL required environment variables at once
-    const apiKey = (0, config_1.getThreadlineApiKey)();
-    const account = (0, config_1.getThreadlineAccount)();
-    const missingVars = [];
-    // Check for undefined, empty string, or literal unexpanded variable
-    // GitLab CI keeps variables as literal "$VAR" if not defined in CI/CD settings
-    if (!apiKey || apiKey === '$THREADLINE_API_KEY')
-        missingVars.push('THREADLINE_API_KEY');
-    if (!account || account === '$THREADLINE_ACCOUNT')
-        missingVars.push('THREADLINE_ACCOUNT');
-    if (missingVars.length > 0) {
-        logger_1.logger.error('Missing required environment variables:');
-        for (const varName of missingVars) {
-            logger_1.logger.error(`   • ${varName}`);
-        }
-        console.log('');
-        console.log(chalk_1.default.yellow('To fix this:'));
-        console.log('');
-        console.log(chalk_1.default.white('  Local development:'));
-        console.log(chalk_1.default.gray('    1. Create a .env.local file in your project root'));
-        console.log(chalk_1.default.gray('    2. Add the missing variable(s):'));
-        if (missingVars.includes('THREADLINE_API_KEY')) {
-            console.log(chalk_1.default.gray('       THREADLINE_API_KEY=your-api-key-here'));
-        }
-        if (missingVars.includes('THREADLINE_ACCOUNT')) {
-            console.log(chalk_1.default.gray('       THREADLINE_ACCOUNT=your-email@example.com'));
-        }
-        console.log(chalk_1.default.gray('    3. Make sure .env.local is in your .gitignore'));
-        console.log('');
-        console.log(chalk_1.default.white('  CI/CD:'));
-        console.log(chalk_1.default.gray('    GitHub Actions:     Settings → Secrets → Add variables'));
-        console.log(chalk_1.default.gray('    GitLab CI:          Settings → CI/CD → Variables'));
-        console.log(chalk_1.default.gray('    Bitbucket Pipelines: Repository settings → Repository variables'));
-        console.log(chalk_1.default.gray('    Vercel:             Settings → Environment Variables'));
-        console.log('');
-        console.log(chalk_1.default.gray('Get your credentials at: https://devthreadline.com/settings'));
+    // Pre-flight check: Validate OpenAI API key is set (required for local processing)
+    const openAIConfig = (0, config_1.getOpenAIConfig)(config);
+    if (!openAIConfig) {
+        logger_1.logger.error('Missing required environment variable: OPENAI_API_KEY');
+        logger_1.logger.output('');
+        logger_1.logger.output(chalk_1.default.yellow('To fix this:'));
+        logger_1.logger.output('');
+        logger_1.logger.output(chalk_1.default.white('  Local development:'));
+        logger_1.logger.output(chalk_1.default.gray('    1. Create a .env.local file in your project root'));
+        logger_1.logger.output(chalk_1.default.gray('    2. Add: OPENAI_API_KEY=your-openai-api-key'));
+        logger_1.logger.output(chalk_1.default.gray('    3. Make sure .env.local is in your .gitignore'));
+        logger_1.logger.output('');
+        logger_1.logger.output(chalk_1.default.white('  CI/CD:'));
+        logger_1.logger.output(chalk_1.default.gray('    GitHub Actions:     Settings → Secrets → Add OPENAI_API_KEY'));
+        logger_1.logger.output(chalk_1.default.gray('    GitLab CI:          Settings → CI/CD → Variables → Add OPENAI_API_KEY'));
+        logger_1.logger.output(chalk_1.default.gray('    Bitbucket Pipelines: Repository settings → Repository variables → Add OPENAI_API_KEY'));
+        logger_1.logger.output(chalk_1.default.gray('    Vercel:             Settings → Environment Variables → Add OPENAI_API_KEY'));
+        logger_1.logger.output('');
+        logger_1.logger.output(chalk_1.default.gray('Get your OpenAI API key at: https://platform.openai.com/api-keys'));
         process.exit(1);
     }
+    // Log OpenAI configuration
+    (0, config_1.logOpenAIConfig)(openAIConfig);
     // 1. Find and validate threadlines
     logger_1.logger.info('Finding threadlines...');
     const threadlines = await (0, experts_1.findThreadlines)(cwd, gitRoot);
-    console.log(chalk_1.default.green(`✓ Found ${threadlines.length} threadline(s)\n`));
+    logger_1.logger.info(`✓ Found ${threadlines.length} threadline(s)\n`);
     if (threadlines.length === 0) {
-        console.log(chalk_1.default.yellow('⚠️  No valid threadlines found.'));
-        console.log(chalk_1.default.gray('   Run `npx threadlines init` to create your first threadline.'));
+        logger_1.logger.warn('No valid threadlines found.');
+        logger_1.logger.output(chalk_1.default.gray('   Run `npx threadlines init` to create your first threadline.'));
         process.exit(0);
     }
     // 2. Detect environment and context
@@ -154,7 +140,7 @@ async function checkCommand(options) {
     // Validate mutually exclusive flags
     if (explicitFlags.length > 1) {
         logger_1.logger.error('Only one review option can be specified at a time');
-        console.log(chalk_1.default.gray('   Options: --commit, --file, --folder, --files'));
+        logger_1.logger.output(chalk_1.default.gray('   Options: --commit, --file, --folder, --files'));
         process.exit(1);
     }
     // CI environments: auto-detect only, flags are ignored with warning
@@ -224,29 +210,29 @@ async function checkCommand(options) {
         }
     }
     if (gitDiff.changedFiles.length === 0) {
-        console.error(chalk_1.default.bold('ℹ️ No changes detected.'));
+        logger_1.logger.info('ℹ️ No changes detected.');
         process.exit(0);
     }
     // Safety limit: prevent expensive API calls on large diffs
     const MAX_CHANGED_FILES = 20;
     if (gitDiff.changedFiles.length > MAX_CHANGED_FILES) {
-        console.error(chalk_1.default.red(`❌ Too many changed files: ${gitDiff.changedFiles.length} (max: ${MAX_CHANGED_FILES})`));
-        console.error(chalk_1.default.gray('   This limit prevents expensive API calls on large diffs.'));
-        console.error(chalk_1.default.gray('   Consider reviewing smaller batches of changes.'));
+        logger_1.logger.error(`Too many changed files: ${gitDiff.changedFiles.length} (max: ${MAX_CHANGED_FILES})`);
+        logger_1.logger.output(chalk_1.default.gray('   This limit prevents expensive API calls on large diffs.'));
+        logger_1.logger.output(chalk_1.default.gray('   Consider reviewing smaller batches of changes.'));
         process.exit(1);
     }
     // Check for zero diff (files changed but no actual code changes)
     if (!gitDiff.diff || gitDiff.diff.trim() === '') {
-        console.log(chalk_1.default.blue('ℹ️  No code changes detected. Diff contains zero lines added or removed.'));
-        console.log(chalk_1.default.gray(`   ${gitDiff.changedFiles.length} file(s) changed but no content modifications detected.`));
-        console.log('');
-        console.log(chalk_1.default.bold('Results:\n'));
-        console.log(chalk_1.default.gray(`${threadlines.length} threadlines checked`));
-        console.log(chalk_1.default.gray(`  ${threadlines.length} not relevant`));
-        console.log('');
+        logger_1.logger.info('ℹ️  No code changes detected. Diff contains zero lines added or removed.');
+        logger_1.logger.output(chalk_1.default.gray(`   ${gitDiff.changedFiles.length} file(s) changed but no content modifications detected.`));
+        logger_1.logger.output('');
+        logger_1.logger.output(chalk_1.default.bold('Results:\n'));
+        logger_1.logger.output(chalk_1.default.gray(`${threadlines.length} threadlines checked`));
+        logger_1.logger.output(chalk_1.default.gray(`  ${threadlines.length} not relevant`));
+        logger_1.logger.output('');
         process.exit(0);
     }
-    console.log(chalk_1.default.green(`✓ Found ${gitDiff.changedFiles.length} changed file(s) (context: ${reviewContext})\n`));
+    logger_1.logger.info(`✓ Found ${gitDiff.changedFiles.length} changed file(s) (context: ${reviewContext})\n`);
     // Log the files being sent
     for (const file of gitDiff.changedFiles) {
         logger_1.logger.info(`  → ${file}`);
@@ -281,26 +267,86 @@ async function checkCommand(options) {
             contextContent
         };
     });
-    // 5. Call review API
+    // 5. Process threadlines locally using OpenAI
     logger_1.logger.info('Running threadline checks...');
-    const client = new client_1.ReviewAPIClient(config.api_url);
-    const response = await client.review({
-        threadlines: threadlinesWithContext,
+    const processResponse = await (0, expert_1.processThreadlines)({
+        threadlines: threadlinesWithContext.map(t => ({
+            id: t.id,
+            version: t.version,
+            patterns: t.patterns,
+            content: t.content,
+            contextFiles: t.contextFiles,
+            contextContent: t.contextContent
+        })),
         diff: gitDiff.diff,
         files: gitDiff.changedFiles,
-        apiKey: apiKey,
-        account: account,
-        repoName: repoName,
-        branchName: branchName,
-        commitSha: metadata.commitSha,
-        commitMessage: metadata.commitMessage,
-        commitAuthorName: metadata.commitAuthorName,
-        commitAuthorEmail: metadata.commitAuthorEmail,
-        prTitle: metadata.prTitle,
-        environment: environment,
-        cliVersion: CLI_VERSION,
-        reviewContext: reviewContext
+        apiKey: openAIConfig.apiKey,
+        model: openAIConfig.model,
+        serviceTier: openAIConfig.serviceTier,
+        contextLinesForLLM: config.diff_context_lines
     });
+    // Convert ProcessThreadlinesResponse to ReviewResponse format for displayResults
+    const response = {
+        results: processResponse.results,
+        metadata: {
+            totalThreadlines: processResponse.metadata.totalThreadlines,
+            completed: processResponse.metadata.completed,
+            timedOut: processResponse.metadata.timedOut,
+            errors: processResponse.metadata.errors
+        }
+    };
+    // 6. Sync results to web app (if mode is "online")
+    if (config.mode === 'online') {
+        const apiKey = (0, config_1.getThreadlineApiKey)();
+        const account = (0, config_1.getThreadlineAccount)();
+        if (!apiKey || !account) {
+            // Configuration error: mode is "online" but credentials are missing
+            // Fail loudly - this is a user configuration error that needs to be fixed
+            logger_1.logger.error('Sync mode is "online" but required credentials are missing.');
+            logger_1.logger.error('Set THREADLINE_API_KEY and THREADLINE_ACCOUNT environment variables to enable syncing.');
+            logger_1.logger.error('Alternatively, set mode to "offline" in .threadlinerc to disable syncing.');
+            throw new Error('Sync configuration error: mode is "online" but THREADLINE_API_KEY or THREADLINE_ACCOUNT is not set. ' +
+                'Either set these environment variables or change mode to "offline" in .threadlinerc.');
+        }
+        // Attempt sync - if it fails, show error but don't fail the check (local processing succeeded)
+        try {
+            logger_1.logger.info('Syncing results to web app...');
+            const client = new client_1.ReviewAPIClient(config.api_url);
+            await client.syncResults({
+                threadlines: threadlinesWithContext,
+                diff: gitDiff.diff,
+                files: gitDiff.changedFiles,
+                results: processResponse.results,
+                metadata: {
+                    totalThreadlines: processResponse.metadata.totalThreadlines,
+                    completed: processResponse.metadata.completed,
+                    timedOut: processResponse.metadata.timedOut,
+                    errors: processResponse.metadata.errors,
+                    llmModel: processResponse.metadata.llmModel
+                },
+                apiKey,
+                account,
+                repoName,
+                branchName,
+                commitSha: metadata.commitSha,
+                commitMessage: metadata.commitMessage,
+                commitAuthorName: metadata.commitAuthorName,
+                commitAuthorEmail: metadata.commitAuthorEmail,
+                prTitle: metadata.prTitle,
+                environment: environment,
+                cliVersion: CLI_VERSION,
+                reviewContext: reviewContext
+            });
+            logger_1.logger.info('✓ Results synced successfully');
+        }
+        catch (error) {
+            // Sync API call failed - show error prominently but don't fail the check (local processing succeeded)
+            // This is not a silent fallback: we explicitly show the error and explain why we continue
+            const errorMessage = error instanceof Error ? error.message : 'Unknown error';
+            logger_1.logger.error(`Failed to sync results to web app: ${errorMessage}`);
+            logger_1.logger.warn('Check results are still valid - sync failure does not affect local processing.');
+        }
+    }
     // 7. Display results (with filtering if --full not specified)
     displayResults(response, options.full || false);
     // Exit with appropriate code (attention or errors = failure)
@@ -315,7 +361,7 @@ function displayResults(response, showFull) {
         : results.filter((r) => r.status === 'attention');
     // Display informational message if present (e.g., zero diffs)
     if (message) {
-        console.log('\n' + chalk_1.default.blue('ℹ️  ' + message));
+        logger_1.logger.output('\n' + chalk_1.default.blue('ℹ️  ' + message));
     }
     const notRelevant = results.filter((r) => r.status === 'not_relevant').length;
     const compliant = results.filter((r) => r.status === 'compliant').length;
@@ -344,81 +390,81 @@ function displayResults(response, showFull) {
     // Show success message with breakdown if no issues
     if (attention === 0 && metadata.timedOut === 0 && errors === 0) {
         const summary = summaryParts.length > 0 ? ` (${summaryParts.join(', ')})` : '';
-        console.log('\n' + chalk_1.default.green(`✓ Threadline check passed${summary}`));
-        console.log(chalk_1.default.gray(`  ${metadata.totalThreadlines} threadline${metadata.totalThreadlines !== 1 ? 's' : ''} checked\n`));
+        logger_1.logger.output('\n' + chalk_1.default.green(`✓ Threadline check passed${summary}`));
+        logger_1.logger.output(chalk_1.default.gray(`  ${metadata.totalThreadlines} threadline${metadata.totalThreadlines !== 1 ? 's' : ''} checked\n`));
     }
     else {
         // Show detailed breakdown when there are issues
-        console.log('\n' + chalk_1.default.bold('Results:\n'));
-        console.log(chalk_1.default.gray(`${metadata.totalThreadlines} threadlines checked`));
+        logger_1.logger.output('\n' + chalk_1.default.bold('Results:\n'));
+        logger_1.logger.output(chalk_1.default.gray(`${metadata.totalThreadlines} threadlines checked`));
         if (showFull) {
             // Show all results when --full flag is used
             if (notRelevant > 0) {
-                console.log(chalk_1.default.gray(`  ${notRelevant} not relevant`));
+                logger_1.logger.output(chalk_1.default.gray(`  ${notRelevant} not relevant`));
             }
             if (compliant > 0) {
-                console.log(chalk_1.default.green(`  ${compliant} compliant`));
+                logger_1.logger.output(chalk_1.default.green(`  ${compliant} compliant`));
             }
             if (attention > 0) {
-                console.log(chalk_1.default.yellow(`  ${attention} attention`));
+                logger_1.logger.output(chalk_1.default.yellow(`  ${attention} attention`));
             }
         }
         else {
             // Default: only show attention items
             if (attention > 0) {
-                console.log(chalk_1.default.yellow(`  ${attention} attention`));
+                logger_1.logger.output(chalk_1.default.yellow(`  ${attention} attention`));
             }
         }
         if (metadata.timedOut > 0) {
-            console.log(chalk_1.default.yellow(`  ${metadata.timedOut} timed out`));
+            logger_1.logger.output(chalk_1.default.yellow(`  ${metadata.timedOut} timed out`));
         }
         if (errors > 0) {
-            console.log(chalk_1.default.red(`  ${errors} errors`));
+            logger_1.logger.output(chalk_1.default.red(`  ${errors} errors`));
         }
-        console.log('');
+        logger_1.logger.output('');
     }
     // Show attention items
     if (attentionItems.length > 0) {
         for (const item of attentionItems) {
-            console.log(chalk_1.default.yellow(`[attention] ${item.expertId}`));
+            logger_1.logger.output(chalk_1.default.yellow(`[attention] ${item.expertId}`));
             if (item.fileReferences && item.fileReferences.length > 0) {
                 // List all files as bullet points
                 for (const fileRef of item.fileReferences) {
                     const lineRef = item.lineReferences?.[item.fileReferences.indexOf(fileRef)];
                     const lineStr = lineRef ? `:${lineRef}` : '';
-                    console.log(chalk_1.default.gray(`* ${fileRef}${lineStr}`));
+                    logger_1.logger.output(chalk_1.default.gray(`* ${fileRef}${lineStr}`));
                 }
             }
             // Show reasoning once at the end (if available)
             if (item.reasoning) {
-                console.log(chalk_1.default.gray(item.reasoning));
+                logger_1.logger.output(chalk_1.default.gray(item.reasoning));
             }
             else if (!item.fileReferences || item.fileReferences.length === 0) {
-                console.log(chalk_1.default.gray('Needs attention'));
+                logger_1.logger.output(chalk_1.default.gray('Needs attention'));
             }
-            console.log(''); // Empty line between threadlines
+            logger_1.logger.output(''); // Empty line between threadlines
         }
     }
     // Show error items (always shown, regardless of --full flag)
     if (errorItems.length > 0) {
         for (const item of errorItems) {
-            console.log(chalk_1.default.red(`[error] ${item.expertId}`));
+            logger_1.logger.output(chalk_1.default.red(`[error] ${item.expertId}`));
             // Show error message
             if (item.error) {
-                console.log(chalk_1.default.red(`  Error: ${item.error.message}`));
+                logger_1.logger.output(chalk_1.default.red(`  Error: ${item.error.message}`));
                 if (item.error.type) {
-                    console.log(chalk_1.default.red(`  Type: ${item.error.type}`));
+                    logger_1.logger.output(chalk_1.default.red(`  Type: ${item.error.type}`));
                 }
                 if (item.error.code) {
-                    console.log(chalk_1.default.red(`  Code: ${item.error.code}`));
+                    logger_1.logger.output(chalk_1.default.red(`  Code: ${item.error.code}`));
                 }
                 // Show raw response for debugging
                 if (item.error.rawResponse) {
-                    console.log(chalk_1.default.gray('  Raw response:'));
-                    console.log(chalk_1.default.gray(JSON.stringify(item.error.rawResponse, null, 2).split('\n').map(line => '    ' + line).join('\n')));
+                    logger_1.logger.output(chalk_1.default.gray('  Raw response:'));
+                    logger_1.logger.output(chalk_1.default.gray(JSON.stringify(item.error.rawResponse, null, 2).split('\n').map(line => '    ' + line).join('\n')));
                 }
             }
-            console.log(''); // Empty line between errors
+            logger_1.logger.output(''); // Empty line between errors
         }
     }
 }