parse_error_log

Instructions

Implementation Reference

• src/tools/log_trimmer.ts:8-93 (handler)

  Core handler function `trimErrorLog` that parses and compresses error logs/stack traces. Extracts error types/messages, filters out node_modules frames, keeps only business code frames up to maxFrames, and returns a compressed summary with stats.

  export function trimErrorLog(
    log: string,
    workspaceRoot?: string,
    maxFrames: number = 10
  ): LogTrimResult {
    const lines = log.split("\n");
    const originalLines = lines.length;
  
    // 1. 提取错误类型和消息
    const errorPattern = /^(\w*Error|\w*Exception|FATAL|PANIC)[:\s]+(.*)$/;
    const errorMatches: Array<{ type: string; message: string; line: number }> = [];
    lines.forEach((line, i) => {
      const match = line.match(errorPattern);
      if (match) errorMatches.push({ type: match[1], message: match[2], line: i });
    });
  
    // 2. 提取堆栈帧，过滤噪音
    const framePattern = /^\s+at\s+(.+?)\s+\((.+?):(\d+):(\d+)\)$/;
    const alternativeFramePattern = /^\s+at\s+(.+?)$/;
    const nodeModulesPattern = /node_modules/;
    const workspacePattern = workspaceRoot
      ? new RegExp(workspaceRoot.replace(/[.*+?^${}()|[\]\\]/g, "\\$&"))
      : /\/src\//;
  
    const businessFrames: string[] = [];
    let totalFrames = 0;
  
    for (const line of lines) {
      const match = line.match(framePattern);
      if (match) {
        totalFrames++;
        const [, func, file] = match;
        if (!nodeModulesPattern.test(file)) {
          businessFrames.push(line.trim());
        }
        continue;
      }
      const altMatch = line.match(alternativeFramePattern);
      if (altMatch) {
        totalFrames++;
        businessFrames.push(line.trim());
      }
    }
  
    // 3. 组装摘要
    const parts: string[] = [];
  
    if (errorMatches.length > 0) {
      const primary = errorMatches[0];
      parts.push(`[${primary.type}] ${primary.message}`);
      if (errorMatches.length > 1) {
        const secondary = errorMatches[1];
        parts.push(`Caused by: [${secondary.type}] ${secondary.message}`);
      }
    }
  
    if (businessFrames.length > 0) {
      parts.push("\nRelevant stack frames:");
      businessFrames.slice(0, maxFrames).forEach((f) => parts.push(`  ${f}`));
    }
  
    // 4. 添加尾部摘要
    const omittedFrames = totalFrames - Math.min(businessFrames.length, maxFrames);
    if (omittedFrames > 0) {
      parts.push(`  ... ${omittedFrames} frames omitted`);
    }
  
    // 如果没有提取到任何有用信息，返回截断的原始日志
    if (parts.length === 0) {
      const maxLogLines = 50;
      const truncated = lines.slice(0, maxLogLines).join("\n");
      return {
        summary: truncated + (lines.length > maxLogLines ? `\n... [${lines.length - maxLogLines} more lines omitted by ContextGC]` : ""),
        stats: { originalLines, summaryLines: Math.min(lines.length, maxLogLines) },
      };
    }
  
    const summary = parts.join("\n");
    return {
      summary,
      stats: {
        originalLines,
        summaryLines: summary.split("\n").length,
      },
    };
  }

• src/tools/registry.ts:153-177 (registration)

  Registration of the 'parse_error_log' tool on the MCP server using `server.tool()`. Defines the schema (logContent, workspaceRoot, maxFrames), description, and calls `trimErrorLog` as the handler.

  // ─────────────────────────────────────────────────────
  // Tool 3: parse_error_log
  // ─────────────────────────────────────────────────────
  server.tool(
    "parse_error_log",
    "Parses and compresses error logs / stack traces to extract only the essential error information. Filters out node_modules frames, keeps only your source code references. Use this instead of reading raw stderr output to save 90%+ tokens.",
    {
      logContent: z.string().describe("The raw error log / stderr output to compress"),
      workspaceRoot: z.string().optional().describe("Absolute path of the workspace root, used to identify business code frames."),
      maxFrames: z.number().optional().describe("Maximum number of stack frames to include. Default: 10."),
    },
    async (args): Promise<ToolResult> => {
      const result = trimErrorLog(
        args.logContent,
        args.workspaceRoot,
        args.maxFrames ?? config.logTrimmer.maxFrames
      );
      return {
        content: [{
          type: "text",
          text: result.summary + `\n// [ContextGC] ${result.stats.originalLines} → ${result.stats.summaryLines} lines`,
        }],
      };
    }
  );

• src/tools/registry.ts:159-163 (schema)

  Zod schema for the tool's input parameters: logContent (required string), workspaceRoot (optional string), maxFrames (optional number).

  {
    logContent: z.string().describe("The raw error log / stderr output to compress"),
    workspaceRoot: z.string().optional().describe("Absolute path of the workspace root, used to identify business code frames."),
    maxFrames: z.number().optional().describe("Maximum number of stack frames to include. Default: 10."),
  },

• src/config/schema.ts:20-23 (helper)

  Type definition for `logTrimmer` config, including maxFrames and filterPatterns.

  logTrimmer: {
    maxFrames: number;
    filterPatterns: string[];
  };

• src/config/schema.ts:45-48 (helper)

  Default configuration values for logTrimmer: maxFrames=10, filterPatterns=['node_modules', '.next', '.cache', 'dist', 'build'].

  logTrimmer: {
    maxFrames: 10,
    filterPatterns: ["node_modules", ".next", ".cache", "dist", "build"],
  },