@mmnto/cli 1.14.12 → 1.14.13

package/dist/commands/run-compiled-rules.js

@@ -9,319 +9,323 @@ export async function runCompiledRules(options) {
     const path = await import('node:path');
     const { bold, errorColor, log, success: successColor } = await import('../ui.js');
     const { writeOutput } = await import('../utils.js');
-    const { applyAstRulesToAdditions, applyRulesToAdditions, enrichWithAstContext, extractAddedLines, loadCompiledRules, loadRuleMetrics, matchesGlob, recordContextHit, recordEvaluation, recordSuppression, recordTrigger, resolveGitRoot, safeExec, saveRuleMetrics, setCoreLogger, TotemError, } = await import('@mmnto/totem');
+    const { applyAstRulesToAdditions, applyRulesToAdditions, enrichWithAstContext, extractAddedLines, loadCompiledRules, loadRuleMetrics, matchesGlob, recordContextHit, recordEvaluation, recordSuppression, recordTrigger, resolveGitRoot, safeExec, saveRuleMetrics, TotemError, } = await import('@mmnto/totem');
     const { diff, cwd, totemDir, format, outPath, exportPaths, ignorePatterns, tag, isStaged } = options;
-    // Wire core logger to CLI UI (ADR-071: core must not use console.warn directly)
-    setCoreLogger({ warn: (msg) => log.warn(tag, msg) });
+    // Per-invocation rule-engine context (ADR-071 + mmnto/totem#1441): logger
+    // threads into the engine as a parameter rather than a module-level setter,
+    // so concurrent / federated runs cannot clobber each other's wiring.
+    const ruleCtx = {
+        logger: { warn: (msg) => log.warn(tag, msg) },
+        state: { hasWarnedShieldContext: false },
+    };
+    const resolvedTotemDir = path.join(options.configRoot ?? cwd, totemDir);
+    // Load compiled rules
+    const rulesPath = path.join(resolvedTotemDir, COMPILED_RULES_FILE);
+    const rules = loadCompiledRules(rulesPath);
+    if (rules.length === 0) {
+        throw new TotemError('NO_RULES', `No compiled rules found at ${totemDir}/${COMPILED_RULES_FILE}.`, "Run 'totem compile' to generate rules.");
+    }
+    log.info(tag, `Running ${rules.length} rules (zero LLM)...`);
+    // Extract additions, exclude compiled rules file, export targets, and binary files
+    const BINARY_EXTENSIONS = new Set([
+        '.png',
+        '.jpg',
+        '.jpeg',
+        '.gif',
+        '.mp4',
+        '.pdf',
+        '.zip',
+        '.tar',
+        '.gz',
+        '.woff',
+        '.woff2',
+        '.eot',
+        '.ttf',
+        '.mp3',
+        '.wav',
+        '.ico',
+        '.bin',
+    ]);
+    const rulesRelPath = path.join(totemDir, COMPILED_RULES_FILE).replace(/\\/g, '/');
+    const excluded = new Set([rulesRelPath]);
+    if (exportPaths) {
+        for (const ep of exportPaths) {
+            excluded.add(ep.replace(/\\/g, '/'));
+        }
+    }
+    const additions = extractAddedLines(diff)
+        .filter((a) => !excluded.has(a.file))
+        .filter((a) => !BINARY_EXTENSIONS.has(path.extname(a.file).toLowerCase()))
+        .filter((a) => !ignorePatterns || !ignorePatterns.some((pattern) => matchesGlob(a.file, pattern)));
+    // Resolve repo root once — git diff paths are always repo-root-relative,
+    // so both staged and non-staged paths need the repo root for file resolution.
+    const repoRoot = resolveGitRoot(cwd);
+    // Enrich with AST context
     try {
-        const resolvedTotemDir = path.join(options.configRoot ?? cwd, totemDir);
-        // Load compiled rules
-        const rulesPath = path.join(resolvedTotemDir, COMPILED_RULES_FILE);
-        const rules = loadCompiledRules(rulesPath);
-        if (rules.length === 0) {
-            throw new TotemError('NO_RULES', `No compiled rules found at ${totemDir}/${COMPILED_RULES_FILE}.`, "Run 'totem compile' to generate rules.");
+        await enrichWithAstContext(additions, { cwd: repoRoot ?? cwd });
+        const classified = additions.filter((a) => a.astContext !== undefined).length;
+        if (classified > 0) {
+            log.dim(tag, `AST classified ${classified}/${additions.length} additions`);
         }
-        log.info(tag, `Running ${rules.length} rules (zero LLM)...`);
-        // Extract additions, exclude compiled rules file, export targets, and binary files
-        const BINARY_EXTENSIONS = new Set([
-            '.png',
-            '.jpg',
-            '.jpeg',
-            '.gif',
-            '.mp4',
-            '.pdf',
-            '.zip',
-            '.tar',
-            '.gz',
-            '.woff',
-            '.woff2',
-            '.eot',
-            '.ttf',
-            '.mp3',
-            '.wav',
-            '.ico',
-            '.bin',
-        ]);
-        const rulesRelPath = path.join(totemDir, COMPILED_RULES_FILE).replace(/\\/g, '/');
-        const excluded = new Set([rulesRelPath]);
-        if (exportPaths) {
-            for (const ep of exportPaths) {
-                excluded.add(ep.replace(/\\/g, '/'));
-            }
+        // totem-context: intentional graceful degradation — AST enrichment is best-effort
+    }
+    catch {
+        log.dim(tag, 'AST classification unavailable, falling back to raw matching');
+    }
+    // Record metrics + Trap Ledger
+    const { appendLedgerEvent } = await import('@mmnto/totem');
+    const metrics = loadRuleMetrics(resolvedTotemDir, (msg) => log.dim(tag, msg));
+    const ruleEventCallback = (event, hash, context) => {
+        if (event === 'trigger') {
+            recordTrigger(metrics, hash);
+            recordContextHit(metrics, hash, context?.astContext);
         }
-        const additions = extractAddedLines(diff)
-            .filter((a) => !excluded.has(a.file))
-            .filter((a) => !BINARY_EXTENSIONS.has(path.extname(a.file).toLowerCase()))
-            .filter((a) => !ignorePatterns || !ignorePatterns.some((pattern) => matchesGlob(a.file, pattern)));
-        // Resolve repo root once — git diff paths are always repo-root-relative,
-        // so both staged and non-staged paths need the repo root for file resolution.
-        const repoRoot = resolveGitRoot(cwd);
-        // Enrich with AST context
-        try {
-            await enrichWithAstContext(additions, { cwd: repoRoot ?? cwd });
-            const classified = additions.filter((a) => a.astContext !== undefined).length;
-            if (classified > 0) {
-                log.dim(tag, `AST classified ${classified}/${additions.length} additions`);
+        else if (event === 'suppress') {
+            recordSuppression(metrics, hash);
+            // Append to Trap Ledger (fire-and-forget). When the suppressed rule
+            // was shipped by a pack with immutable: true (ADR-089,
+            // mmnto-ai/totem#1485), the event carries the flag so auditors can
+            // surface every attempt to silence an enforced security rule via
+            // `jq 'select(.immutable == true)'` over events.ndjson.
+            if (context) {
+                appendLedgerEvent(resolvedTotemDir, {
+                    timestamp: new Date().toISOString(),
+                    type: context.justification ? 'override' : 'suppress',
+                    ruleId: hash,
+                    file: context.file,
+                    line: context.line,
+                    justification: context.justification ?? '',
+                    source: 'lint',
+                    ...(context.immutable === true ? { immutable: true } : {}),
+                }, (msg) => log.dim(tag, msg));
             }
         }
-        catch {
-            log.dim(tag, 'AST classification unavailable, falling back to raw matching');
+        else {
+            // mmnto/totem#1408: 'failure' event fires when a compiled rule's
+            // runtime findAll throws (per-rule try/catch in executeQuery). Log
+            // the hash and reason so the operator can see WHICH rule failed
+            // without crashing the batch. Metric recording (recordFailure) is
+            // a follow-up once `rule-metrics` gains a failure counter.
+            log.warn(tag, `rule ${hash} failed at runtime${context?.failureReason ? `: ${context.failureReason}` : ''}`);
         }
-        // Record metrics + Trap Ledger
-        const { appendLedgerEvent } = await import('@mmnto/totem');
-        const metrics = loadRuleMetrics(resolvedTotemDir, (msg) => log.dim(tag, msg));
-        const ruleEventCallback = (event, hash, context) => {
-            if (event === 'trigger') {
-                recordTrigger(metrics, hash);
-                recordContextHit(metrics, hash, context?.astContext);
-            }
-            else if (event === 'suppress') {
-                recordSuppression(metrics, hash);
-                // Append to Trap Ledger (fire-and-forget). When the suppressed rule
-                // was shipped by a pack with immutable: true (ADR-089,
-                // mmnto-ai/totem#1485), the event carries the flag so auditors can
-                // surface every attempt to silence an enforced security rule via
-                // `jq 'select(.immutable == true)'` over events.ndjson.
-                if (context) {
-                    appendLedgerEvent(resolvedTotemDir, {
-                        timestamp: new Date().toISOString(),
-                        type: context.justification ? 'override' : 'suppress',
-                        ruleId: hash,
-                        file: context.file,
-                        line: context.line,
-                        justification: context.justification ?? '',
-                        source: 'lint',
-                        ...(context.immutable === true ? { immutable: true } : {}),
-                    }, (msg) => log.dim(tag, msg));
-                }
-            }
-            else {
-                // mmnto/totem#1408: 'failure' event fires when a compiled rule's
-                // runtime findAll throws (per-rule try/catch in executeQuery). Log
-                // the hash and reason so the operator can see WHICH rule failed
-                // without crashing the batch. Metric recording (recordFailure) is
-                // a follow-up once `rule-metrics` gains a failure counter.
-                log.warn(tag, `rule ${hash} failed at runtime${context?.failureReason ? `: ${context.failureReason}` : ''}`);
-            }
-        };
-        const regexViolations = applyRulesToAdditions(rules, additions, ruleEventCallback);
-        // Run AST rules (async — reads files and runs Tree-sitter/ast-grep queries)
-        const astRules = rules.filter((r) => r.engine === 'ast' || r.engine === 'ast-grep');
-        let astViolations = [];
-        if (astRules.length > 0) {
-            log.dim(tag, `Running ${astRules.length} AST rule(s)...`);
-            try {
-                const workingDirectory = repoRoot ?? cwd;
-                let readStrategy = undefined;
-                if (isStaged) {
-                    if (repoRoot) {
-                        readStrategy = async (filePath) => {
-                            try {
-                                // 1. Detect symlinks explicitly (git ls-files -s returns mode 120000).
-                                //    The `--` separator prevents filePath values starting with `-` from
-                                //    being interpreted as git options.
-                                const lsOutput = safeExec('git', ['ls-files', '--recurse-submodules', '-s', '--', filePath], { cwd: repoRoot, env: { ...process.env, LC_ALL: 'C' } });
-                                if (lsOutput.startsWith('120000 ')) {
-                                    return null; // Explicitly exclude symlinks from AST checks
-                                }
-                                // 2. Read staged content
-                                const content = safeExec('git', ['show', `:${filePath}`], {
-                                    cwd: repoRoot,
-                                    trim: false,
-                                    env: { ...process.env, LC_ALL: 'C' },
-                                });
-                                // 3. Normalize CRLF to LF specifically for the staged callback
-                                // Disk-read callback preserves existing behavior per Invariant #4.
-                                return content.replace(/\r\n/g, '\n');
-                            }
-                            catch (err) {
-                                // Explicit throw per Failure Mode 1 decision
-                                throw new TotemError('STAGED_READ_FAILED', `Failed to read staged content for ${filePath}`, `git show :${filePath} failed. The file may not exist in the index or may be staged for deletion. Ensure --staged is used correctly.`, { cause: err });
+    };
+    const regexViolations = applyRulesToAdditions(ruleCtx, rules, additions, ruleEventCallback);
+    // Run AST rules (async — reads files and runs Tree-sitter/ast-grep queries)
+    const astRules = rules.filter((r) => r.engine === 'ast' || r.engine === 'ast-grep');
+    let astViolations = [];
+    if (astRules.length > 0) {
+        log.dim(tag, `Running ${astRules.length} AST rule(s)...`);
+        try {
+            const workingDirectory = repoRoot ?? cwd;
+            let readStrategy = undefined;
+            if (isStaged) {
+                if (repoRoot) {
+                    readStrategy = async (filePath) => {
+                        try {
+                            // totem-context: false positive — comment mentions `git ls-files`; the actual call below already uses --recurse-submodules
+                            // 1. Detect symlinks explicitly (git ls-files -s returns mode 120000).
+                            //    The `--` separator prevents filePath values starting with `-` from
+                            //    being interpreted as git options.
+                            const lsOutput = safeExec('git', ['ls-files', '--recurse-submodules', '-s', '--', filePath], { cwd: repoRoot, env: { ...process.env, LC_ALL: 'C' } });
+                            if (lsOutput.startsWith('120000 ')) {
+                                return null; // Explicitly exclude symlinks from AST checks
                             }
-                        };
-                    }
-                }
-                astViolations = await applyAstRulesToAdditions(rules, additions, workingDirectory, ruleEventCallback, (msg) => log.warn(tag, msg), readStrategy);
-            }
-            catch (err) {
-                // STAGED_READ_FAILED must propagate — the pre-commit guarantee depends
-                // on surfacing staged-read failures rather than silently falling back.
-                if (err instanceof TotemError && err.code === 'STAGED_READ_FAILED') {
-                    throw err;
-                }
-                const msg = err instanceof Error ? err.message : String(err);
-                const isWasmFailure = /not initialized|wasm|web-tree-sitter/i.test(msg);
-                if (process.env['TOTEM_LITE'] === '1' && isWasmFailure) {
-                    // In the lite binary, WASM init may fail under Node.js (works in Bun).
-                    // Degrade gracefully: skip AST rules, warn, continue with regex results.
-                    log.warn(tag, `AST rules skipped (WASM engine unavailable): ${msg}`);
-                }
-                else {
-                    throw err;
+                            // 2. Read staged content
+                            const content = safeExec('git', ['show', `:${filePath}`], {
+                                cwd: repoRoot,
+                                trim: false,
+                                env: { ...process.env, LC_ALL: 'C' },
+                            });
+                            // 3. Normalize CRLF to LF specifically for the staged callback
+                            // totem-context: Invariant #4 refers to an internal invariant number, not an issue ref
+                            // Disk-read callback preserves existing behavior per Invariant #4.
+                            return content.replace(/\r\n/g, '\n');
+                        }
+                        catch (err) {
+                            // Explicit throw per Failure Mode 1 decision
+                            throw new TotemError('STAGED_READ_FAILED', `Failed to read staged content for ${filePath}`, `git show :${filePath} failed. The file may not exist in the index or may be staged for deletion. Ensure --staged is used correctly.`, { cause: err });
+                        }
+                    };
                 }
             }
+            astViolations = await applyAstRulesToAdditions(ruleCtx, rules, additions, workingDirectory, ruleEventCallback, (msg) => log.warn(tag, msg), readStrategy);
         }
-        const violations = [...regexViolations, ...astViolations];
-        // ── Zero-match rule detection (#1061) ────────────
-        // Count rules whose fileGlobs matched none of the files in this diff.
-        const diffFiles = [...new Set(additions.map((a) => a.file))];
-        const zeroMatchRules = [];
-        for (const rule of rules) {
-            if (rule.fileGlobs && rule.fileGlobs.length > 0) {
-                const positive = rule.fileGlobs.filter((g) => typeof g === 'string' && !g.startsWith('!'));
-                const negative = rule.fileGlobs
-                    .filter((g) => typeof g === 'string' && g.startsWith('!'))
-                    .map((g) => g.slice(1));
-                const hasMatch = diffFiles.some((file) => {
-                    const positiveMatch = positive.length === 0 || positive.some((g) => matchesGlob(file, g));
-                    const negativeMatch = negative.some((g) => matchesGlob(file, g));
-                    return positiveMatch && !negativeMatch;
-                });
-                if (!hasMatch)
-                    zeroMatchRules.push(rule);
+        catch (err) {
+            // STAGED_READ_FAILED must propagate — the pre-commit guarantee depends
+            // on surfacing staged-read failures rather than silently falling back.
+            if (err instanceof TotemError && err.code === 'STAGED_READ_FAILED') {
+                throw err;
+            }
+            const msg = err instanceof Error ? err.message : String(err);
+            const isWasmFailure = /not initialized|wasm|web-tree-sitter/i.test(msg);
+            if (process.env['TOTEM_LITE'] === '1' && isWasmFailure) {
+                // In the lite binary, WASM init may fail under Node.js (works in Bun).
+                // Degrade gracefully: skip AST rules, warn, continue with regex results.
+                log.warn(tag, `AST rules skipped (WASM engine unavailable): ${msg}`);
+            }
+            else {
+                throw err;
             }
         }
-        if (zeroMatchRules.length > 0) {
-            log.dim(tag, `${zeroMatchRules.length} rule(s) matched no files in this diff`);
-        }
-        // mmnto-ai/totem#1483: tick evaluationCount once per rule per lint run.
-        // Invariant: one run loads the rule set, evaluates each rule against the
-        // diff additions, and increments the counter here exactly once per
-        // lessonHash. Multiple matches on a rule within one run still produce a
-        // single increment. This counter is the "was this rule exercised" signal
-        // the doctor stale-rule check reads to distinguish a dormant rule from a
-        // rule that has genuinely sat through N lint cycles without firing.
-        for (const rule of rules) {
-            recordEvaluation(metrics, rule.lessonHash);
-        }
-        try {
-            saveRuleMetrics(resolvedTotemDir, metrics);
-        }
-        catch (err) {
-            log.warn(tag, `Could not save rule metrics: ${err instanceof Error ? err.message : String(err)}`);
+    }
+    const violations = [...regexViolations, ...astViolations];
+    // ── Zero-match rule detection (mmnto-ai/totem#1061) ────────────
+    // Count rules whose fileGlobs matched none of the files in this diff.
+    const diffFiles = [...new Set(additions.map((a) => a.file))];
+    const zeroMatchRules = [];
+    for (const rule of rules) {
+        if (rule.fileGlobs && rule.fileGlobs.length > 0) {
+            const positive = rule.fileGlobs.filter((g) => typeof g === 'string' && !g.startsWith('!'));
+            const negative = rule.fileGlobs
+                .filter((g) => typeof g === 'string' && g.startsWith('!'))
+                .map((g) => g.slice(1));
+            const hasMatch = diffFiles.some((file) => {
+                const positiveMatch = positive.length === 0 || positive.some((g) => matchesGlob(file, g));
+                const negativeMatch = negative.some((g) => matchesGlob(file, g));
+                return positiveMatch && !negativeMatch;
+            });
+            if (!hasMatch)
+                zeroMatchRules.push(rule);
         }
-        // Classify violations by severity (computed once, reused across all output formats)
-        const errors = violations.filter((v) => (v.rule.severity ?? 'error') === 'error');
-        const warnings = violations.filter((v) => (v.rule.severity ?? 'error') === 'warning');
-        // Convert to unified findings model once (ADR-071)
-        const { violationToFinding } = await import('@mmnto/totem');
-        const findings = violations.map(violationToFinding);
-        // Build output
-        let output;
-        if (format === 'sarif') {
-            const { buildSarifLog, getHeadSha } = await import('@mmnto/totem');
-            const { createRequire } = await import('node:module');
-            const req = createRequire(import.meta.url);
-            const version = req('../../package.json').version;
-            const commitHash = getHeadSha(cwd) ?? undefined;
-            // SARIF is a strict channel for error-severity findings only.
-            // Warnings are probationary (Rule Nursery) and stay as local telemetry
-            // to prevent alert fatigue in the PR UI (Proposal 190).
-            const sarif = buildSarifLog(errors, rules, { version, commitHash });
-            // Surface warning count as a single note so users know they exist
-            if (warnings.length > 0) {
-                const summaryRuleIdx = sarif.runs[0].tool.driver.rules.length;
-                sarif.runs[0].tool.driver.rules.push({
-                    id: 'totem/warning-summary',
-                    shortDescription: {
-                        text: 'Probationary warnings detected — run `totem lint` locally to review',
-                    },
-                });
-                sarif.runs[0].results.push({
-                    ruleId: 'totem/warning-summary',
-                    ruleIndex: summaryRuleIdx,
-                    level: 'note',
-                    message: {
-                        text: `${warnings.length} warning-severity finding(s) detected. Warnings are probationary and not shown in PR reviews. Run \`totem lint\` locally to review.`,
-                    },
-                    locations: [
-                        {
-                            physicalLocation: {
-                                artifactLocation: { uri: '.totem/compiled-rules.json' },
-                                region: { startLine: 1 },
-                            },
+    }
+    if (zeroMatchRules.length > 0) {
+        log.dim(tag, `${zeroMatchRules.length} rule(s) matched no files in this diff`);
+    }
+    // mmnto-ai/totem#1483: tick evaluationCount once per rule per lint run.
+    // Invariant: one run loads the rule set, evaluates each rule against the
+    // diff additions, and increments the counter here exactly once per
+    // lessonHash. Multiple matches on a rule within one run still produce a
+    // single increment. This counter is the "was this rule exercised" signal
+    // the doctor stale-rule check reads to distinguish a dormant rule from a
+    // rule that has genuinely sat through N lint cycles without firing.
+    for (const rule of rules) {
+        recordEvaluation(metrics, rule.lessonHash);
+    }
+    try {
+        saveRuleMetrics(resolvedTotemDir, metrics);
+        // totem-context: intentional graceful degradation — metric save is best-effort
+    }
+    catch (err) {
+        log.warn(tag, `Could not save rule metrics: ${err instanceof Error ? err.message : String(err)}`);
+    }
+    // Classify violations by severity (computed once, reused across all output formats)
+    const errors = violations.filter((v) => (v.rule.severity ?? 'error') === 'error');
+    const warnings = violations.filter((v) => (v.rule.severity ?? 'error') === 'warning');
+    // Convert to unified findings model once (ADR-071)
+    const { violationToFinding } = await import('@mmnto/totem');
+    const findings = violations.map(violationToFinding);
+    // Build output
+    let output;
+    if (format === 'sarif') {
+        const { buildSarifLog, getHeadSha } = await import('@mmnto/totem');
+        const { createRequire } = await import('node:module');
+        const req = createRequire(import.meta.url);
+        const version = req('../../package.json').version;
+        const commitHash = getHeadSha(cwd) ?? undefined;
+        // SARIF is a strict channel for error-severity findings only.
+        // Warnings are probationary (Rule Nursery) and stay as local telemetry
+        // to prevent alert fatigue in the PR UI (Proposal 190).
+        const sarif = buildSarifLog(errors, rules, { version, commitHash });
+        // Surface warning count as a single note so users know they exist
+        if (warnings.length > 0) {
+            const summaryRuleIdx = sarif.runs[0].tool.driver.rules.length;
+            sarif.runs[0].tool.driver.rules.push({
+                id: 'totem/warning-summary',
+                shortDescription: {
+                    text: 'Probationary warnings detected — run `totem lint` locally to review',
+                },
+            });
+            sarif.runs[0].results.push({
+                ruleId: 'totem/warning-summary',
+                ruleIndex: summaryRuleIdx,
+                level: 'note',
+                message: {
+                    text: `${warnings.length} warning-severity finding(s) detected. Warnings are probationary and not shown in PR reviews. Run \`totem lint\` locally to review.`,
+                },
+                locations: [
+                    {
+                        physicalLocation: {
+                            artifactLocation: { uri: '.totem/compiled-rules.json' },
+                            region: { startLine: 1 },
                         },
-                    ],
-                });
-            }
-            output = JSON.stringify(sarif, null, 2);
+                    },
+                ],
+            });
         }
-        else if (format === 'json') {
-            output = JSON.stringify({
-                pass: errors.length === 0,
-                rules: rules.length,
-                errors: errors.length,
-                warnings: warnings.length,
-                findings,
-                violations,
-            }, null, 2);
+        output = JSON.stringify(sarif, null, 2);
+    }
+    else if (format === 'json') {
+        output = JSON.stringify({
+            pass: errors.length === 0,
+            rules: rules.length,
+            errors: errors.length,
+            warnings: warnings.length,
+            findings,
+            violations,
+        }, null, 2);
+    }
+    else {
+        const lines = [];
+        if (errors.length === 0 && warnings.length === 0) {
+            // Clean pass — only emit verbose markdown when writing to file
+            if (outPath) {
+                lines.push('### Verdict');
+                lines.push(`**PASS** - All ${rules.length} rules passed.`);
+                lines.push('');
+                lines.push('### Details');
+                lines.push('No violations detected against compiled rules.');
+            }
         }
         else {
-            const lines = [];
-            if (errors.length === 0 && warnings.length === 0) {
-                // Clean pass — only emit verbose markdown when writing to file
-                if (outPath) {
-                    lines.push('### Verdict');
-                    lines.push(`**PASS** - All ${rules.length} rules passed.`);
-                    lines.push('');
-                    lines.push('### Details');
-                    lines.push('No violations detected against compiled rules.');
-                }
+            lines.push('### Verdict');
+            if (errors.length > 0) {
+                lines.push(`**FAIL** - ${errors.length} error(s)${warnings.length > 0 ? `, ${warnings.length} warning(s)` : ''} across ${rules.length} rules.`);
             }
             else {
-                lines.push('### Verdict');
-                if (errors.length > 0) {
-                    lines.push(`**FAIL** - ${errors.length} error(s)${warnings.length > 0 ? `, ${warnings.length} warning(s)` : ''} across ${rules.length} rules.`);
-                }
-                else {
-                    lines.push(`**PASS** - ${warnings.length} warning(s), 0 errors across ${rules.length} rules.`);
-                }
-                if (errors.length > 0) {
+                lines.push(`**PASS** - ${warnings.length} warning(s), 0 errors across ${rules.length} rules.`);
+            }
+            if (errors.length > 0) {
+                lines.push('');
+                lines.push('### Errors');
+                for (const v of errors) {
+                    lines.push(`- **${v.file}:${v.lineNumber}** - ${v.rule.message}`);
+                    lines.push(`  Pattern: \`/${v.rule.pattern}/\``);
+                    lines.push(`  Lesson: "${v.rule.lessonHeading}"`);
+                    lines.push(`  Line: \`${v.line.trim()}\``);
                     lines.push('');
-                    lines.push('### Errors');
-                    for (const v of errors) {
-                        lines.push(`- **${v.file}:${v.lineNumber}** - ${v.rule.message}`);
-                        lines.push(`  Pattern: \`/${v.rule.pattern}/\``);
-                        lines.push(`  Lesson: "${v.rule.lessonHeading}"`);
-                        lines.push(`  Line: \`${v.line.trim()}\``);
-                        lines.push('');
-                    }
                 }
-                if (warnings.length > 0) {
+            }
+            if (warnings.length > 0) {
+                lines.push('');
+                lines.push('### Warnings');
+                for (const v of warnings) {
+                    lines.push(`- **${v.file}:${v.lineNumber}** - ${v.rule.message}`);
+                    lines.push(`  Pattern: \`/${v.rule.pattern}/\``);
+                    lines.push(`  Lesson: "${v.rule.lessonHeading}"`);
+                    lines.push(`  Line: \`${v.line.trim()}\``);
                     lines.push('');
-                    lines.push('### Warnings');
-                    for (const v of warnings) {
-                        lines.push(`- **${v.file}:${v.lineNumber}** - ${v.rule.message}`);
-                        lines.push(`  Pattern: \`/${v.rule.pattern}/\``);
-                        lines.push(`  Lesson: "${v.rule.lessonHeading}"`);
-                        lines.push(`  Line: \`${v.line.trim()}\``);
-                        lines.push('');
-                    }
                 }
             }
-            output = lines.join('\n');
-        }
-        writeOutput(output, outPath);
-        if (outPath)
-            log.success(tag, `Written to ${outPath}`);
-        if (errors.length > 0) {
-            const verdictLabel = errorColor(bold('FAIL'));
-            const warnSuffix = warnings.length > 0 ? `, ${warnings.length} warning(s)` : '';
-            log.info(tag, `Verdict: ${verdictLabel} - ${errors.length} error(s)${warnSuffix}`);
-            throw new TotemError('SHIELD_FAILED', 'Violations detected', 'Fix the violations above or use totem explain <hash> for details.');
         }
-        else if (warnings.length > 0) {
-            const verdictLabel = successColor(bold('PASS'));
-            log.info(tag, `Verdict: ${verdictLabel} - ${warnings.length} warning(s), 0 errors`);
-        }
-        else {
-            const verdictLabel = successColor(bold('PASS'));
-            log.info(tag, `Verdict: ${verdictLabel} - ${rules.length} rules, 0 violations`);
-        }
-        return { violations, findings, rules, output };
+        output = lines.join('\n');
+    }
+    writeOutput(output, outPath);
+    if (outPath)
+        log.success(tag, `Written to ${outPath}`);
+    if (errors.length > 0) {
+        const verdictLabel = errorColor(bold('FAIL'));
+        const warnSuffix = warnings.length > 0 ? `, ${warnings.length} warning(s)` : '';
+        log.info(tag, `Verdict: ${verdictLabel} - ${errors.length} error(s)${warnSuffix}`);
+        throw new TotemError('SHIELD_FAILED', 'Violations detected', 'Fix the violations above or use totem explain <hash> for details.');
+    }
+    else if (warnings.length > 0) {
+        const verdictLabel = successColor(bold('PASS'));
+        log.info(tag, `Verdict: ${verdictLabel} - ${warnings.length} warning(s), 0 errors`);
     }
-    finally {
-        setCoreLogger({ warn: () => { } });
+    else {
+        const verdictLabel = successColor(bold('PASS'));
+        log.info(tag, `Verdict: ${verdictLabel} - ${rules.length} rules, 0 violations`);
     }
+    return { violations, findings, rules, output };
 }
 //# sourceMappingURL=run-compiled-rules.js.map