agentsys 5.0.3 → 5.1.0

package/plugins/debate/lib/enhance/cross-file-patterns.js

@@ -0,0 +1,370 @@
+/**
+ * Cross-File Semantic Analysis Patterns
+ * Detection patterns for multi-file consistency and alignment
+ *
+ * Cross-platform compatible: Works with Claude Code, OpenCode, and Codex
+ *
+ * @author Avi Fenesh
+ * @license MIT
+ */
+
+const fs = require('fs');
+const path = require('path');
+const { getStateDir, getPlatformName } = require('../platform/state-dir');
+
+/**
+ * Platform-specific default tools
+ * These are used when no tools.json config is found
+ */
+const PLATFORM_TOOLS = {
+  claude: [
+    'Task', 'Read', 'Write', 'Edit', 'Glob', 'Grep', 'Bash',
+    'WebFetch', 'WebSearch', 'AskUserQuestion', 'NotebookEdit',
+    'LSP', 'Skill', 'EnterPlanMode', 'ExitPlanMode',
+    'TaskCreate', 'TaskUpdate', 'TaskList', 'TaskGet', 'TaskOutput', 'TaskStop'
+  ],
+  opencode: [
+    'Task', 'Read', 'Write', 'Edit', 'Glob', 'Grep', 'Bash',
+    'WebFetch', 'WebSearch', 'AskUser', 'Notebook',
+    'LSP', 'Skill', 'Plan'
+  ],
+  codex: [
+    'Task', 'Read', 'Write', 'Edit', 'Glob', 'Grep', 'Bash',
+    'WebFetch', 'Ask', 'Shell'
+  ],
+  // Superset for unknown platforms
+  unknown: [
+    'Task', 'Read', 'Write', 'Edit', 'Glob', 'Grep', 'Bash',
+    'WebFetch', 'WebSearch', 'AskUserQuestion', 'AskUser', 'Ask',
+    'NotebookEdit', 'Notebook', 'LSP', 'Skill', 'Plan',
+    'EnterPlanMode', 'ExitPlanMode', 'Shell',
+    'TaskCreate', 'TaskUpdate', 'TaskList', 'TaskGet', 'TaskOutput', 'TaskStop'
+  ]
+};
+
+/** Maximum size for tools.json config file (prevents DoS via large files) */
+const MAX_CONFIG_SIZE = 64 * 1024; // 64KB
+
+/**
+ * Load tools configuration from state directory or use platform defaults
+ * @param {string} [basePath=process.cwd()] - Base path to check
+ * @returns {string[]} Array of known tool names
+ */
+function loadKnownTools(basePath = process.cwd()) {
+  const stateDir = getStateDir(basePath);
+  const toolsConfigPath = path.join(basePath, stateDir, 'tools.json');
+
+  // Try to load from config file (without race condition)
+  try {
+    const content = fs.readFileSync(toolsConfigPath, 'utf8');
+
+    // Size limit check
+    if (content.length > MAX_CONFIG_SIZE) {
+      if (process.env.DEBUG) {
+        console.error('[cross-file-patterns] tools.json exceeds size limit, using defaults');
+      }
+      throw new Error('Config file too large');
+    }
+
+    const config = JSON.parse(content);
+
+    // Validate config structure
+    if (Array.isArray(config.tools)) {
+      return config.tools;
+    }
+    if (config.knownTools && Array.isArray(config.knownTools)) {
+      return config.knownTools;
+    }
+
+    if (process.env.DEBUG) {
+      console.error('[cross-file-patterns] tools.json missing tools array, using defaults');
+    }
+  } catch (err) {
+    // ENOENT (file not found) is expected, don't log
+    // Other errors should be logged for debugging
+    if (err.code !== 'ENOENT' && process.env.DEBUG) {
+      console.error('[cross-file-patterns] Failed to load tools.json:', err.message);
+    }
+  }
+
+  // Fall back to platform-specific defaults
+  const platform = getPlatformName(basePath);
+  return PLATFORM_TOOLS[platform] || PLATFORM_TOOLS.unknown;
+}
+
+/**
+ * Cross-file semantic analysis patterns
+ * All patterns use MEDIUM certainty since cross-file analysis requires human review
+ */
+const crossFilePatterns = {
+  /**
+   * Tool mentioned in prompt body not in frontmatter tools list
+   */
+  tool_not_in_allowed_list: {
+    id: 'tool_not_in_allowed_list',
+    category: 'tool-consistency',
+    certainty: 'MEDIUM',
+    autoFix: false,
+    description: 'Prompt mentions tools not declared in frontmatter',
+    check: (data) => {
+      const { declaredTools, usedTools, agentName } = data;
+      if (!declaredTools || declaredTools.length === 0) return null;
+      if (!usedTools || usedTools.length === 0) return null;
+
+      // Normalize declared tools (handle Bash(git:*) -> Bash)
+      const normalizedDeclared = declaredTools.map(t => t.split('(')[0].trim());
+
+      const undeclared = usedTools.filter(tool =>
+        !normalizedDeclared.includes(tool) &&
+        !normalizedDeclared.includes('*')
+      );
+
+      if (undeclared.length > 0) {
+        return {
+          issue: `Agent "${agentName}" uses ${undeclared.join(', ')} but not declared in tools frontmatter`,
+          fix: `Add ${undeclared.join(', ')} to tools field or remove usage from prompt`
+        };
+      }
+      return null;
+    }
+  },
+
+  /**
+   * Workflow references an agent that does not exist
+   */
+  missing_workflow_agent: {
+    id: 'missing_workflow_agent',
+    category: 'workflow',
+    certainty: 'MEDIUM',
+    autoFix: false,
+    description: 'Workflow references non-existent agent',
+    check: (data) => {
+      const { referencedAgent, existingAgents, sourceFile } = data;
+      if (!referencedAgent || !existingAgents) return null;
+
+      // Extract plugin:agent format
+      const [plugin, agentName] = referencedAgent.includes(':')
+        ? referencedAgent.split(':')
+        : [null, referencedAgent];
+
+      // Check if agent exists
+      const exists = existingAgents.some(a => {
+        if (plugin) {
+          return a.plugin === plugin && a.name === agentName;
+        }
+        return a.name === agentName;
+      });
+
+      if (!exists) {
+        return {
+          issue: `Referenced agent "${referencedAgent}" does not exist`,
+          fix: `Create agent "${referencedAgent}" or fix the reference in ${sourceFile}`
+        };
+      }
+      return null;
+    }
+  },
+
+  /**
+   * Workflow phases not fully connected
+   */
+  incomplete_phase_transition: {
+    id: 'incomplete_phase_transition',
+    category: 'workflow',
+    certainty: 'MEDIUM',
+    autoFix: false,
+    description: 'Workflow phase transitions are incomplete',
+    check: (data) => {
+      const { phases, transitions, workflowFile } = data;
+      if (!phases || !transitions || phases.length === 0) return null;
+
+      // Check each phase has a transition (except last)
+      const missingTransitions = [];
+      for (let i = 0; i < phases.length - 1; i++) {
+        const currentPhase = phases[i];
+        const hasTransition = transitions.some(t =>
+          t.from === currentPhase || t.to === phases[i + 1]
+        );
+        if (!hasTransition) {
+          missingTransitions.push(`${currentPhase} -> ${phases[i + 1]}`);
+        }
+      }
+
+      if (missingTransitions.length > 0) {
+        return {
+          issue: `Missing phase transitions: ${missingTransitions.join(', ')}`,
+          fix: `Add transition logic for ${missingTransitions[0]} in ${workflowFile}`
+        };
+      }
+      return null;
+    }
+  },
+
+  /**
+   * Duplicate instructions across multiple agents
+   */
+  duplicate_instructions: {
+    id: 'duplicate_instructions',
+    category: 'consistency',
+    certainty: 'MEDIUM',
+    autoFix: false,
+    description: 'Identical critical instructions across agents',
+    check: (data) => {
+      const { instruction, files } = data;
+      if (!instruction || !files || files.length < 2) return null;
+
+      return {
+        issue: `Duplicate instruction found in ${files.length} files: "${instruction.substring(0, 50)}..."`,
+        fix: `Extract shared instruction to a common include or ensure intentional duplication`
+      };
+    }
+  },
+
+  /**
+   * Contradictory rules across agents
+   */
+  contradictory_rules: {
+    id: 'contradictory_rules',
+    category: 'consistency',
+    certainty: 'MEDIUM',
+    autoFix: false,
+    description: 'Conflicting rules across agents (always vs never)',
+    check: (data) => {
+      const { rule1, rule2, file1, file2 } = data;
+      if (!rule1 || !rule2) return null;
+
+      return {
+        issue: `Contradictory rules: "${rule1.substring(0, 40)}..." vs "${rule2.substring(0, 40)}..."`,
+        fix: `Resolve conflict between ${file1} and ${file2}`
+      };
+    }
+  },
+
+  /**
+   * Prompt not referenced by any workflow or skill
+   */
+  orphaned_prompt: {
+    id: 'orphaned_prompt',
+    category: 'consistency',
+    certainty: 'MEDIUM',
+    autoFix: false,
+    description: 'Agent/prompt not referenced anywhere',
+    check: (data) => {
+      const { promptFile, referencedBy } = data;
+      if (!promptFile) return null;
+
+      if (!referencedBy || referencedBy.length === 0) {
+        return {
+          issue: `Orphaned prompt: ${promptFile} is not referenced by any workflow or skill`,
+          fix: `Add reference to ${promptFile} or remove if unused`
+        };
+      }
+      return null;
+    }
+  },
+
+  /**
+   * Skill allowed-tools differs from what prompt uses
+   */
+  skill_tool_mismatch: {
+    id: 'skill_tool_mismatch',
+    category: 'skill-alignment',
+    certainty: 'MEDIUM',
+    autoFix: false,
+    description: 'Skill allowed-tools differs from prompt tool usage',
+    check: (data) => {
+      const { skillName, skillAllowedTools, promptUsedTools } = data;
+      if (!skillAllowedTools || !promptUsedTools) return null;
+
+      // Normalize skill tools
+      const normalizedSkill = skillAllowedTools.map(t => t.split('(')[0].trim());
+
+      const mismatches = promptUsedTools.filter(tool =>
+        !normalizedSkill.includes(tool) &&
+        !normalizedSkill.includes('*')
+      );
+
+      if (mismatches.length > 0) {
+        return {
+          issue: `Skill "${skillName}" prompt uses ${mismatches.join(', ')} not in allowed-tools`,
+          fix: `Add ${mismatches.join(', ')} to skill allowed-tools or update prompt`
+        };
+      }
+      return null;
+    }
+  },
+
+  /**
+   * Skill description does not match actual behavior
+   */
+  skill_description_mismatch: {
+    id: 'skill_description_mismatch',
+    category: 'skill-alignment',
+    certainty: 'MEDIUM',
+    autoFix: false,
+    description: 'Skill description mentions capabilities not found in body',
+    check: (data) => {
+      const { skillName, description, bodyCapabilities, descriptionCapabilities } = data;
+      if (!description || !bodyCapabilities) return null;
+
+      const missingInBody = descriptionCapabilities.filter(cap =>
+        !bodyCapabilities.some(bc => bc.toLowerCase().includes(cap.toLowerCase()))
+      );
+
+      if (missingInBody.length > 0) {
+        return {
+          issue: `Skill "${skillName}" description mentions "${missingInBody[0]}" not found in body`,
+          fix: `Update description to match actual capabilities or add missing functionality`
+        };
+      }
+      return null;
+    }
+  }
+};
+
+/**
+ * Get all cross-file patterns
+ * @returns {Object} All patterns
+ */
+function getAllPatterns() {
+  return crossFilePatterns;
+}
+
+/**
+ * Get patterns by category
+ * @param {string} category - tool-consistency, workflow, consistency, skill-alignment
+ * @returns {Object} Filtered patterns
+ */
+function getPatternsByCategory(category) {
+  const result = {};
+  for (const [name, pattern] of Object.entries(crossFilePatterns)) {
+    if (pattern.category === category) {
+      result[name] = pattern;
+    }
+  }
+  return result;
+}
+
+/**
+ * Get patterns by certainty
+ * @param {string} certainty - HIGH, MEDIUM, LOW
+ * @returns {Object} Filtered patterns
+ */
+function getPatternsByCertainty(certainty) {
+  const result = {};
+  for (const [name, pattern] of Object.entries(crossFilePatterns)) {
+    if (pattern.certainty === certainty) {
+      result[name] = pattern;
+    }
+  }
+  return result;
+}
+
+module.exports = {
+  PLATFORM_TOOLS,
+  loadKnownTools,
+  crossFilePatterns,
+  getAllPatterns,
+  getPatternsByCategory,
+  getPatternsByCertainty
+};

package/plugins/debate/lib/enhance/docs-analyzer.js

@@ -0,0 +1,325 @@
+/**
+ * Documentation Analyzer
+ * @author Avi Fenesh
+ * @license MIT
+ */
+
+const fs = require('fs');
+const path = require('path');
+const { getPatternsForMode, estimateTokens } = require('./docs-patterns');
+
+function analyzeDoc(docPath, options = {}) {
+  const { mode = 'both', verbose = false, existingFiles = [] } = options;
+
+  const results = {
+    docName: path.basename(docPath, '.md'),
+    docPath,
+    mode,
+    tokenCount: 0,
+    linkIssues: [],
+    structureIssues: [],
+    codeIssues: [],
+    efficiencyIssues: [],
+    ragIssues: [],
+    balanceIssues: []
+  };
+
+  // Read file
+  if (!fs.existsSync(docPath)) {
+    results.structureIssues.push({
+      issue: 'File not found',
+      file: docPath,
+      certainty: 'HIGH',
+      patternId: 'file_not_found'
+    });
+    return results;
+  }
+
+  let content;
+  try {
+    content = fs.readFileSync(docPath, 'utf8');
+  } catch (err) {
+    results.structureIssues.push({
+      issue: `Failed to read file: ${err.message}`,
+      file: docPath,
+      certainty: 'HIGH',
+      patternId: 'read_error'
+    });
+    return results;
+  }
+
+  // Calculate token count
+  results.tokenCount = estimateTokens(content);
+
+  // Get patterns applicable to this mode
+  const patterns = getPatternsForMode(mode);
+
+  // Context for pattern checks
+  const context = { existingFiles };
+
+  // Run each pattern check
+  for (const [patternName, pattern] of Object.entries(patterns)) {
+    // Skip LOW certainty unless verbose
+    if (pattern.certainty === 'LOW' && !verbose) {
+      continue;
+    }
+
+    // Run the check
+    const result = pattern.check(content, context);
+
+    if (result) {
+      const issue = {
+        ...result,
+        file: docPath,
+        certainty: pattern.certainty,
+        patternId: pattern.id,
+        autoFix: pattern.autoFix
+      };
+
+      // Route to appropriate issue category
+      switch (pattern.category) {
+        case 'link':
+          results.linkIssues.push(issue);
+          break;
+        case 'structure':
+          results.structureIssues.push(issue);
+          break;
+        case 'code':
+          results.codeIssues.push(issue);
+          break;
+        case 'efficiency':
+          results.efficiencyIssues.push(issue);
+          break;
+        case 'rag':
+          results.ragIssues.push(issue);
+          break;
+        case 'balance':
+          results.balanceIssues.push(issue);
+          break;
+        default:
+          results.structureIssues.push(issue);
+      }
+    }
+  }
+
+  return results;
+}
+
+function analyzeAllDocs(docsDir, options = {}) {
+  const { recursive = true, ...analyzeOptions } = options;
+  const results = [];
+
+  if (!fs.existsSync(docsDir)) {
+    return results;
+  }
+
+  // Collect all markdown files
+  const mdFiles = [];
+
+  function findMdFiles(dir) {
+    let entries;
+    try {
+      entries = fs.readdirSync(dir, { withFileTypes: true });
+    } catch (err) {
+      return;
+    }
+
+    for (const entry of entries) {
+      const fullPath = path.join(dir, entry.name);
+
+      if (entry.isDirectory() && recursive) {
+        // Skip common non-doc directories
+        if (!['node_modules', '.git', 'dist', 'build'].includes(entry.name)) {
+          findMdFiles(fullPath);
+        }
+      } else if (entry.isFile() && entry.name.endsWith('.md')) {
+        // Skip README files in nested directories for agent-docs mode
+        if (options.mode === 'ai' && entry.name === 'README.md') {
+          continue;
+        }
+        mdFiles.push(fullPath);
+      }
+    }
+  }
+
+  findMdFiles(docsDir);
+
+  // Get relative paths for link validation
+  const existingFiles = mdFiles.map(f => path.relative(docsDir, f).replace(/\\/g, '/'));
+
+  // Analyze each file
+  for (const mdFile of mdFiles) {
+    const result = analyzeDoc(mdFile, { ...analyzeOptions, existingFiles });
+    results.push(result);
+  }
+
+  return results;
+}
+
+function analyze(options = {}) {
+  const {
+    doc,
+    docsDir = 'docs',
+    mode = 'both',
+    verbose = false
+  } = options;
+
+  if (doc) {
+    // Check if doc is a directory or file
+    try {
+      const stats = fs.statSync(doc);
+      if (stats.isDirectory()) {
+        // Analyze all docs in directory
+        return analyzeAllDocs(doc, { mode, verbose });
+      } else {
+        // Analyze single doc
+        return analyzeDoc(doc, { mode, verbose });
+      }
+    } catch (err) {
+      // If file doesn't exist, let analyzeDoc handle the error
+      return analyzeDoc(doc, { mode, verbose });
+    }
+  } else {
+    // Analyze all docs in directory
+    return analyzeAllDocs(docsDir, { mode, verbose });
+  }
+}
+
+function applyFixes(results, options = {}) {
+  // Collect all issues
+  let allIssues = [];
+
+  if (Array.isArray(results)) {
+    for (const r of results) {
+      allIssues.push(...(r.linkIssues || []));
+      allIssues.push(...(r.structureIssues || []));
+      allIssues.push(...(r.codeIssues || []));
+      allIssues.push(...(r.efficiencyIssues || []));
+      allIssues.push(...(r.ragIssues || []));
+      allIssues.push(...(r.balanceIssues || []));
+    }
+  } else {
+    allIssues.push(...(results.linkIssues || []));
+    allIssues.push(...(results.structureIssues || []));
+    allIssues.push(...(results.codeIssues || []));
+    allIssues.push(...(results.efficiencyIssues || []));
+    allIssues.push(...(results.ragIssues || []));
+    allIssues.push(...(results.balanceIssues || []));
+  }
+
+  // Filter to auto-fixable docs issues
+  const docsFixablePatternIds = [
+    'inconsistent_heading_levels',
+    'verbose_explanations'
+  ];
+
+  const fixableIssues = allIssues.filter(i =>
+    i.certainty === 'HIGH' &&
+    i.autoFix &&
+    docsFixablePatternIds.includes(i.patternId)
+  );
+
+  // Apply fixes using the fixer module's pattern
+  return applyDocsFixes(fixableIssues, options);
+}
+
+function applyDocsFixes(issues, options = {}) {
+  const { dryRun = false, backup = true } = options;
+  const fixer = require('./fixer');
+
+  const results = {
+    applied: [],
+    skipped: [],
+    errors: []
+  };
+
+  // Group by file
+  const byFile = new Map();
+  for (const issue of issues) {
+    const fp = issue.file;
+    if (!byFile.has(fp)) {
+      byFile.set(fp, []);
+    }
+    byFile.get(fp).push(issue);
+  }
+
+  // Process each file
+  for (const [filePath, fileIssues] of byFile) {
+    try {
+      if (!fs.existsSync(filePath)) {
+        results.errors.push({ filePath, error: 'File not found' });
+        continue;
+      }
+
+      let content = fs.readFileSync(filePath, 'utf8');
+      const appliedToFile = [];
+
+      for (const issue of fileIssues) {
+        try {
+          if (issue.patternId === 'inconsistent_heading_levels') {
+            content = fixer.fixInconsistentHeadings(content);
+            appliedToFile.push({
+              issue: issue.issue,
+              fix: 'Fixed heading levels',
+              filePath
+            });
+          } else if (issue.patternId === 'verbose_explanations') {
+            content = fixer.fixVerboseExplanations(content);
+            appliedToFile.push({
+              issue: issue.issue,
+              fix: 'Simplified verbose phrases',
+              filePath
+            });
+          }
+        } catch (err) {
+          results.errors.push({
+            issue: issue.issue,
+            filePath,
+            error: err.message
+          });
+        }
+      }
+
+      // Write changes
+      if (!dryRun && appliedToFile.length > 0) {
+        if (backup) {
+          fs.writeFileSync(`${filePath}.backup`, fs.readFileSync(filePath, 'utf8'), 'utf8');
+        }
+        fs.writeFileSync(filePath, content, 'utf8');
+      }
+
+      results.applied.push(...appliedToFile);
+
+    } catch (err) {
+      results.errors.push({
+        filePath,
+        error: err.message
+      });
+    }
+  }
+
+  return results;
+}
+
+function generateReport(results, options = {}) {
+  const reporter = require('./reporter');
+
+  if (Array.isArray(results)) {
+    return reporter.generateDocsSummaryReport(results, options);
+  } else {
+    return reporter.generateDocsReport(results, options);
+  }
+}
+
+const fixer = require('./fixer');
+
+module.exports = {
+  analyzeDoc,
+  analyzeAllDocs,
+  analyze,
+  applyFixes,
+  applyDocsFixes,
+  fixInconsistentHeadings: fixer.fixInconsistentHeadings,
+  fixVerboseExplanations: fixer.fixVerboseExplanations,
+  generateReport
+};