@nerviq/cli 1.20.0 → 1.21.0

package/src/harmony/governance.js

@@ -1,313 +1,313 @@
-/**
- * H8. Harmony Governance - Cross-Platform Policy Enforcement
- *
- * Evaluates cross-platform governance posture: minimum trust levels,
- * required instruction coverage, MCP alignment, and per-platform compliance.
- *
- * Zero external dependencies - imports from sibling/parent modules only.
- */
-
-// ─── Required canonical sections every platform should cover ──────────────────
-
-const REQUIRED_INSTRUCTION_SECTIONS = [
-  { key: 'role', label: 'Role / Identity', description: 'Define who the AI assistant is in this project.' },
-  { key: 'commands', label: 'Verification Commands', description: 'Test, lint, and build commands the AI should use.' },
-  { key: 'language', label: 'Language Preference', description: 'Communication language for the AI.' },
-  { key: 'permissions', label: 'Permission Boundaries', description: 'What the AI is allowed and denied to do.' },
-  { key: 'context', label: 'Project Context', description: 'Architecture, stack, and codebase overview.' },
-  { key: 'safety', label: 'Safety / Secret Protection', description: 'Rules for handling secrets and sensitive files.' },
-];
-
-// ─── Trust posture definitions ────────────────────────────────────────────────
-
-const TRUST_LEVELS = {
-  strict: {
-    level: 1,
-    label: 'Strict',
-    description: 'Read-only or suggest-only. No file writes without approval.',
-    minPermissions: ['deny-write', 'deny-destructive'],
-  },
-  guarded: {
-    level: 2,
-    label: 'Guarded',
-    description: 'Safe writes with explicit deny rules and hook enforcement.',
-    minPermissions: ['deny-destructive', 'secret-protection'],
-  },
-  standard: {
-    level: 3,
-    label: 'Standard',
-    description: 'Accept-edits with common-sense deny rules.',
-    minPermissions: ['secret-protection'],
-  },
-  permissive: {
-    level: 4,
-    label: 'Permissive',
-    description: 'Broad autonomy with minimal restrictions.',
-    minPermissions: [],
-  },
-  unrestricted: {
-    level: 5,
-    label: 'Unrestricted',
-    description: 'Full bypass, only for internal research.',
-    minPermissions: [],
-  },
-};
-
-// ─── Platform trust detection ─────────────────────────────────────────────────
-
-/**
- * Infer trust level from a platform's audit/governance data.
- */
-function inferTrustLevel(platformGovernance) {
-  if (!platformGovernance) return TRUST_LEVELS.unrestricted;
-
-  const { hasPermissions, hasDenyRules, hasSecretProtection, defaultMode } = platformGovernance;
-
-  if (defaultMode === 'plan' || defaultMode === 'read-only') return TRUST_LEVELS.strict;
-  if (hasDenyRules && hasSecretProtection) return TRUST_LEVELS.guarded;
-  if (hasPermissions || hasSecretProtection) return TRUST_LEVELS.standard;
-  if (hasPermissions) return TRUST_LEVELS.permissive;
-  return TRUST_LEVELS.unrestricted;
-}
-
-// ─── Governance evaluation ────────────────────────────────────────────────────
-
-/**
- * Evaluate instruction coverage for a single platform.
- */
-function evaluateInstructionCoverage(platformSections) {
-  const covered = [];
-  const missing = [];
-
-  for (const required of REQUIRED_INSTRUCTION_SECTIONS) {
-    const found = (platformSections || []).some(s => {
-      const key = s.key || s;
-      return key === required.key || key.includes(required.key);
-    });
-    if (found) {
-      covered.push(required);
-    } else {
-      missing.push(required);
-    }
-  }
-
-  return {
-    total: REQUIRED_INSTRUCTION_SECTIONS.length,
-    covered: covered.length,
-    missing: missing.length,
-    coveredSections: covered,
-    missingSections: missing,
-    percentage: Math.round((covered.length / REQUIRED_INSTRUCTION_SECTIONS.length) * 100),
-  };
-}
-
-/**
- * Evaluate MCP server alignment across platforms.
- */
-function evaluateMcpAlignment(platformAudits) {
-  const mcpByPlatform = {};
-  const allServers = new Set();
-
-  for (const audit of (platformAudits || [])) {
-    const servers = audit.mcpServers || [];
-    mcpByPlatform[audit.platform] = new Set(servers);
-    for (const s of servers) allServers.add(s);
-  }
-
-  const alignment = [];
-  for (const server of allServers) {
-    const presentIn = [];
-    const missingFrom = [];
-
-    for (const audit of (platformAudits || [])) {
-      if (mcpByPlatform[audit.platform] && mcpByPlatform[audit.platform].has(server)) {
-        presentIn.push(audit.platform);
-      } else {
-        missingFrom.push(audit.platform);
-      }
-    }
-
-    alignment.push({
-      server,
-      presentIn,
-      missingFrom,
-      aligned: missingFrom.length === 0,
-    });
-  }
-
-  return {
-    totalServers: allServers.size,
-    fullyAligned: alignment.filter(a => a.aligned).length,
-    misaligned: alignment.filter(a => !a.aligned).length,
-    servers: alignment,
-  };
-}
-
-/**
- * Get the complete harmony governance summary.
- *
- * @param {Object|null} canonicalModel - The canonical model from canon.js
- * @param {Array} [platformAudits] - Array of per-platform audit results with governance data
- * @returns {Object} Full governance summary
- */
-function getHarmonyGovernanceSummary(canonicalModel, platformAudits) {
-  const audits = platformAudits || [];
-
-  // 1. Minimum trust posture (lowest acceptable across all platforms)
-  const trustLevels = audits.map(audit => ({
-    platform: audit.platform,
-    trust: inferTrustLevel(audit.governance),
-  }));
-
-  const lowestTrust = trustLevels.length > 0
-    ? trustLevels.reduce((min, curr) =>
-        curr.trust.level < min.trust.level ? curr : min
-      )
-    : null;
-
-  const minimumTrustPosture = {
-    level: lowestTrust ? lowestTrust.trust : TRUST_LEVELS.unrestricted,
-    lowestPlatform: lowestTrust ? lowestTrust.platform : null,
-    perPlatform: trustLevels.map(t => ({
-      platform: t.platform,
-      level: t.trust.label,
-      levelNumber: t.trust.level,
-      description: t.trust.description,
-    })),
-  };
-
-  // 2. Required instruction coverage per platform
-  const instructionCoverage = {
-    required: REQUIRED_INSTRUCTION_SECTIONS,
-    perPlatform: audits.map(audit => ({
-      platform: audit.platform,
-      ...evaluateInstructionCoverage(audit.sections),
-    })),
-  };
-
-  // 3. MCP alignment
-  const mcpAlignment = evaluateMcpAlignment(audits);
-
-  // 4. Compliance: per-platform governance status
-  const platformCompliance = audits.map(audit => {
-    const coverage = evaluateInstructionCoverage(audit.sections);
-    const trust = inferTrustLevel(audit.governance);
-    const gaps = [];
-
-    if (coverage.missing > 0) {
-      gaps.push(`Missing ${coverage.missing} required instruction section(s): ${coverage.missingSections.map(s => s.key).join(', ')}`);
-    }
-
-    if (trust.level >= 4) {
-      gaps.push(`Trust level "${trust.label}" is too permissive for production use`);
-    }
-
-    if (audit.governance && !audit.governance.hasSecretProtection) {
-      gaps.push('No secret protection configured');
-    }
-
-    if (audit.governance && !audit.governance.hasDenyRules) {
-      gaps.push('No deny rules configured');
-    }
-
-    return {
-      platform: audit.platform,
-      compliant: gaps.length === 0,
-      score: audit.score || 0,
-      trustLevel: trust.label,
-      coveragePercent: coverage.percentage,
-      gaps,
-    };
-  });
-
-  return {
-    minimumTrustPosture,
-    instructionCoverage,
-    mcpAlignment,
-    platformCompliance,
-    evaluatedAt: new Date().toISOString(),
-  };
-}
-
-// ─── Report formatting ────────────────────────────────────────────────────────
-
-const COLORS = {
-  reset: '\x1b[0m', bold: '\x1b[1m', dim: '\x1b[2m',
-  green: '\x1b[32m', yellow: '\x1b[33m', red: '\x1b[31m', blue: '\x1b[36m',
-};
-const col = (text, color) => `${COLORS[color] || ''}${text}${COLORS.reset}`;
-
-/**
- * Format the governance summary as a human-readable console report.
- */
-function formatHarmonyGovernanceReport(summary, options) {
-  if (options && options.json) {
-    return JSON.stringify(summary, null, 2);
-  }
-
-  const lines = [];
-
-  lines.push('');
-  lines.push(col('  Harmony Governance Report', 'bold'));
-  lines.push(col('  ═══════════════════════════════════════', 'dim'));
-  lines.push('');
-
-  // Trust posture
-  lines.push(col('  Trust Posture', 'bold'));
-  for (const entry of summary.minimumTrustPosture.perPlatform) {
-    const trustColor = entry.levelNumber <= 2 ? 'green' : entry.levelNumber <= 3 ? 'yellow' : 'red';
-    lines.push(`    ${entry.platform}: ${col(entry.level, trustColor)} — ${entry.description}`);
-  }
-  if (summary.minimumTrustPosture.lowestPlatform) {
-    lines.push(`    ${col('Minimum:', 'dim')} ${summary.minimumTrustPosture.level.label} (${summary.minimumTrustPosture.lowestPlatform})`);
-  }
-  lines.push('');
-
-  // Instruction coverage
-  lines.push(col('  Instruction Coverage', 'bold'));
-  for (const entry of summary.instructionCoverage.perPlatform) {
-    const pctColor = entry.percentage >= 80 ? 'green' : entry.percentage >= 50 ? 'yellow' : 'red';
-    lines.push(`    ${entry.platform}: ${col(`${entry.percentage}%`, pctColor)} (${entry.covered}/${entry.total})`);
-    if (entry.missingSections.length > 0) {
-      lines.push(`      ${col('Missing:', 'dim')} ${entry.missingSections.map(s => s.key).join(', ')}`);
-    }
-  }
-  lines.push('');
-
-  // MCP alignment
-  lines.push(col('  MCP Alignment', 'bold'));
-  lines.push(`    Total servers: ${summary.mcpAlignment.totalServers}`);
-  lines.push(`    Fully aligned: ${col(String(summary.mcpAlignment.fullyAligned), 'green')}`);
-  lines.push(`    Misaligned: ${col(String(summary.mcpAlignment.misaligned), summary.mcpAlignment.misaligned > 0 ? 'yellow' : 'green')}`);
-  for (const server of summary.mcpAlignment.servers) {
-    if (!server.aligned) {
-      lines.push(`      ${col(server.server, 'yellow')}: missing from ${server.missingFrom.join(', ')}`);
-    }
-  }
-  lines.push('');
-
-  // Platform compliance
-  lines.push(col('  Platform Compliance', 'bold'));
-  for (const entry of summary.platformCompliance) {
-    const status = entry.compliant
-      ? col('COMPLIANT', 'green')
-      : col('NON-COMPLIANT', 'red');
-    lines.push(`    ${entry.platform}: ${status} (score: ${entry.score}, trust: ${entry.trustLevel}, coverage: ${entry.coveragePercent}%)`);
-    for (const gap of entry.gaps) {
-      lines.push(`      ${col('GAP:', 'yellow')} ${gap}`);
-    }
-  }
-  lines.push('');
-
-  return lines.join('\n');
-}
-
-module.exports = {
-  getHarmonyGovernanceSummary,
-  formatHarmonyGovernanceReport,
-  inferTrustLevel,
-  evaluateInstructionCoverage,
-  evaluateMcpAlignment,
-  TRUST_LEVELS,
-  REQUIRED_INSTRUCTION_SECTIONS,
-};
+/**
+ * H8. Harmony Governance - Cross-Platform Policy Enforcement
+ *
+ * Evaluates cross-platform governance posture: minimum trust levels,
+ * required instruction coverage, MCP alignment, and per-platform compliance.
+ *
+ * Zero external dependencies - imports from sibling/parent modules only.
+ */
+
+// ─── Required canonical sections every platform should cover ──────────────────
+
+const REQUIRED_INSTRUCTION_SECTIONS = [
+  { key: 'role', label: 'Role / Identity', description: 'Define who the AI assistant is in this project.' },
+  { key: 'commands', label: 'Verification Commands', description: 'Test, lint, and build commands the AI should use.' },
+  { key: 'language', label: 'Language Preference', description: 'Communication language for the AI.' },
+  { key: 'permissions', label: 'Permission Boundaries', description: 'What the AI is allowed and denied to do.' },
+  { key: 'context', label: 'Project Context', description: 'Architecture, stack, and codebase overview.' },
+  { key: 'safety', label: 'Safety / Secret Protection', description: 'Rules for handling secrets and sensitive files.' },
+];
+
+// ─── Trust posture definitions ────────────────────────────────────────────────
+
+const TRUST_LEVELS = {
+  strict: {
+    level: 1,
+    label: 'Strict',
+    description: 'Read-only or suggest-only. No file writes without approval.',
+    minPermissions: ['deny-write', 'deny-destructive'],
+  },
+  guarded: {
+    level: 2,
+    label: 'Guarded',
+    description: 'Safe writes with explicit deny rules and hook enforcement.',
+    minPermissions: ['deny-destructive', 'secret-protection'],
+  },
+  standard: {
+    level: 3,
+    label: 'Standard',
+    description: 'Accept-edits with common-sense deny rules.',
+    minPermissions: ['secret-protection'],
+  },
+  permissive: {
+    level: 4,
+    label: 'Permissive',
+    description: 'Broad autonomy with minimal restrictions.',
+    minPermissions: [],
+  },
+  unrestricted: {
+    level: 5,
+    label: 'Unrestricted',
+    description: 'Full bypass, only for internal research.',
+    minPermissions: [],
+  },
+};
+
+// ─── Platform trust detection ─────────────────────────────────────────────────
+
+/**
+ * Infer trust level from a platform's audit/governance data.
+ */
+function inferTrustLevel(platformGovernance) {
+  if (!platformGovernance) return TRUST_LEVELS.unrestricted;
+
+  const { hasPermissions, hasDenyRules, hasSecretProtection, defaultMode } = platformGovernance;
+
+  if (defaultMode === 'plan' || defaultMode === 'read-only') return TRUST_LEVELS.strict;
+  if (hasDenyRules && hasSecretProtection) return TRUST_LEVELS.guarded;
+  if (hasPermissions || hasSecretProtection) return TRUST_LEVELS.standard;
+  if (hasPermissions) return TRUST_LEVELS.permissive;
+  return TRUST_LEVELS.unrestricted;
+}
+
+// ─── Governance evaluation ────────────────────────────────────────────────────
+
+/**
+ * Evaluate instruction coverage for a single platform.
+ */
+function evaluateInstructionCoverage(platformSections) {
+  const covered = [];
+  const missing = [];
+
+  for (const required of REQUIRED_INSTRUCTION_SECTIONS) {
+    const found = (platformSections || []).some(s => {
+      const key = s.key || s;
+      return key === required.key || key.includes(required.key);
+    });
+    if (found) {
+      covered.push(required);
+    } else {
+      missing.push(required);
+    }
+  }
+
+  return {
+    total: REQUIRED_INSTRUCTION_SECTIONS.length,
+    covered: covered.length,
+    missing: missing.length,
+    coveredSections: covered,
+    missingSections: missing,
+    percentage: Math.round((covered.length / REQUIRED_INSTRUCTION_SECTIONS.length) * 100),
+  };
+}
+
+/**
+ * Evaluate MCP server alignment across platforms.
+ */
+function evaluateMcpAlignment(platformAudits) {
+  const mcpByPlatform = {};
+  const allServers = new Set();
+
+  for (const audit of (platformAudits || [])) {
+    const servers = audit.mcpServers || [];
+    mcpByPlatform[audit.platform] = new Set(servers);
+    for (const s of servers) allServers.add(s);
+  }
+
+  const alignment = [];
+  for (const server of allServers) {
+    const presentIn = [];
+    const missingFrom = [];
+
+    for (const audit of (platformAudits || [])) {
+      if (mcpByPlatform[audit.platform] && mcpByPlatform[audit.platform].has(server)) {
+        presentIn.push(audit.platform);
+      } else {
+        missingFrom.push(audit.platform);
+      }
+    }
+
+    alignment.push({
+      server,
+      presentIn,
+      missingFrom,
+      aligned: missingFrom.length === 0,
+    });
+  }
+
+  return {
+    totalServers: allServers.size,
+    fullyAligned: alignment.filter(a => a.aligned).length,
+    misaligned: alignment.filter(a => !a.aligned).length,
+    servers: alignment,
+  };
+}
+
+/**
+ * Get the complete harmony governance summary.
+ *
+ * @param {Object|null} canonicalModel - The canonical model from canon.js
+ * @param {Array} [platformAudits] - Array of per-platform audit results with governance data
+ * @returns {Object} Full governance summary
+ */
+function getHarmonyGovernanceSummary(canonicalModel, platformAudits) {
+  const audits = platformAudits || [];
+
+  // 1. Minimum trust posture (lowest acceptable across all platforms)
+  const trustLevels = audits.map(audit => ({
+    platform: audit.platform,
+    trust: inferTrustLevel(audit.governance),
+  }));
+
+  const lowestTrust = trustLevels.length > 0
+    ? trustLevels.reduce((min, curr) =>
+        curr.trust.level < min.trust.level ? curr : min
+      )
+    : null;
+
+  const minimumTrustPosture = {
+    level: lowestTrust ? lowestTrust.trust : TRUST_LEVELS.unrestricted,
+    lowestPlatform: lowestTrust ? lowestTrust.platform : null,
+    perPlatform: trustLevels.map(t => ({
+      platform: t.platform,
+      level: t.trust.label,
+      levelNumber: t.trust.level,
+      description: t.trust.description,
+    })),
+  };
+
+  // 2. Required instruction coverage per platform
+  const instructionCoverage = {
+    required: REQUIRED_INSTRUCTION_SECTIONS,
+    perPlatform: audits.map(audit => ({
+      platform: audit.platform,
+      ...evaluateInstructionCoverage(audit.sections),
+    })),
+  };
+
+  // 3. MCP alignment
+  const mcpAlignment = evaluateMcpAlignment(audits);
+
+  // 4. Compliance: per-platform governance status
+  const platformCompliance = audits.map(audit => {
+    const coverage = evaluateInstructionCoverage(audit.sections);
+    const trust = inferTrustLevel(audit.governance);
+    const gaps = [];
+
+    if (coverage.missing > 0) {
+      gaps.push(`Missing ${coverage.missing} required instruction section(s): ${coverage.missingSections.map(s => s.key).join(', ')}`);
+    }
+
+    if (trust.level >= 4) {
+      gaps.push(`Trust level "${trust.label}" is too permissive for production use`);
+    }
+
+    if (audit.governance && !audit.governance.hasSecretProtection) {
+      gaps.push('No secret protection configured');
+    }
+
+    if (audit.governance && !audit.governance.hasDenyRules) {
+      gaps.push('No deny rules configured');
+    }
+
+    return {
+      platform: audit.platform,
+      compliant: gaps.length === 0,
+      score: audit.score || 0,
+      trustLevel: trust.label,
+      coveragePercent: coverage.percentage,
+      gaps,
+    };
+  });
+
+  return {
+    minimumTrustPosture,
+    instructionCoverage,
+    mcpAlignment,
+    platformCompliance,
+    evaluatedAt: new Date().toISOString(),
+  };
+}
+
+// ─── Report formatting ────────────────────────────────────────────────────────
+
+const COLORS = {
+  reset: '\x1b[0m', bold: '\x1b[1m', dim: '\x1b[2m',
+  green: '\x1b[32m', yellow: '\x1b[33m', red: '\x1b[31m', blue: '\x1b[36m',
+};
+const col = (text, color) => `${COLORS[color] || ''}${text}${COLORS.reset}`;
+
+/**
+ * Format the governance summary as a human-readable console report.
+ */
+function formatHarmonyGovernanceReport(summary, options) {
+  if (options && options.json) {
+    return JSON.stringify(summary, null, 2);
+  }
+
+  const lines = [];
+
+  lines.push('');
+  lines.push(col('  Harmony Governance Report', 'bold'));
+  lines.push(col('  ═══════════════════════════════════════', 'dim'));
+  lines.push('');
+
+  // Trust posture
+  lines.push(col('  Trust Posture', 'bold'));
+  for (const entry of summary.minimumTrustPosture.perPlatform) {
+    const trustColor = entry.levelNumber <= 2 ? 'green' : entry.levelNumber <= 3 ? 'yellow' : 'red';
+    lines.push(`    ${entry.platform}: ${col(entry.level, trustColor)} — ${entry.description}`);
+  }
+  if (summary.minimumTrustPosture.lowestPlatform) {
+    lines.push(`    ${col('Minimum:', 'dim')} ${summary.minimumTrustPosture.level.label} (${summary.minimumTrustPosture.lowestPlatform})`);
+  }
+  lines.push('');
+
+  // Instruction coverage
+  lines.push(col('  Instruction Coverage', 'bold'));
+  for (const entry of summary.instructionCoverage.perPlatform) {
+    const pctColor = entry.percentage >= 80 ? 'green' : entry.percentage >= 50 ? 'yellow' : 'red';
+    lines.push(`    ${entry.platform}: ${col(`${entry.percentage}%`, pctColor)} (${entry.covered}/${entry.total})`);
+    if (entry.missingSections.length > 0) {
+      lines.push(`      ${col('Missing:', 'dim')} ${entry.missingSections.map(s => s.key).join(', ')}`);
+    }
+  }
+  lines.push('');
+
+  // MCP alignment
+  lines.push(col('  MCP Alignment', 'bold'));
+  lines.push(`    Total servers: ${summary.mcpAlignment.totalServers}`);
+  lines.push(`    Fully aligned: ${col(String(summary.mcpAlignment.fullyAligned), 'green')}`);
+  lines.push(`    Misaligned: ${col(String(summary.mcpAlignment.misaligned), summary.mcpAlignment.misaligned > 0 ? 'yellow' : 'green')}`);
+  for (const server of summary.mcpAlignment.servers) {
+    if (!server.aligned) {
+      lines.push(`      ${col(server.server, 'yellow')}: missing from ${server.missingFrom.join(', ')}`);
+    }
+  }
+  lines.push('');
+
+  // Platform compliance
+  lines.push(col('  Platform Compliance', 'bold'));
+  for (const entry of summary.platformCompliance) {
+    const status = entry.compliant
+      ? col('COMPLIANT', 'green')
+      : col('NON-COMPLIANT', 'red');
+    lines.push(`    ${entry.platform}: ${status} (score: ${entry.score}, trust: ${entry.trustLevel}, coverage: ${entry.coveragePercent}%)`);
+    for (const gap of entry.gaps) {
+      lines.push(`      ${col('GAP:', 'yellow')} ${gap}`);
+    }
+  }
+  lines.push('');
+
+  return lines.join('\n');
+}
+
+module.exports = {
+  getHarmonyGovernanceSummary,
+  formatHarmonyGovernanceReport,
+  inferTrustLevel,
+  evaluateInstructionCoverage,
+  evaluateMcpAlignment,
+  TRUST_LEVELS,
+  REQUIRED_INSTRUCTION_SECTIONS,
+};