Skip to main content
Glama

cli_tool

Execute CLI commands with configurable sync/async modes, timeout controls, and safety filters for secure automation workflows.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
commandNo要执行的单行命令
commandsNo要执行的多行命令序列 (与 'command' 互斥)
modeNo执行模式: sync (同步阻塞), async (异步非阻塞)sync
timeoutNo命令执行的超时时间(秒)
cwdNo命令执行的工作目录 (绝对路径)
platformNo强制指定执行命令的操作系统环境auto
safe_modeNo是否启用危险命令过滤

Implementation Reference

  • Main execution handler for the 'cli_tool'. Parses arguments, performs validation, safety checks, executes single or multiple CLI commands synchronously or asynchronously using child_process.spawn, handles timeouts and errors.
    export default async function (request: any) {
      try {
        const { command, commands, mode, timeout, cwd, platform: requestedPlatform, safe_mode } = request.params.arguments;
    
        if (!command && !commands) {
          throw new Error("必须提供 'command' 或 'commands' 参数之一。");
        }
    
        if (command && commands) {
          throw new Error("'command' 和 'commands' 参数不能同时提供。");
        }
    
        const osPlatform = requestedPlatform === 'auto' ? platform() : requestedPlatform;
        const options = {
          cwd: cwd || process.cwd(),
          timeout: timeout || 60,
          platform: osPlatform
        };
    
        const allCommands = command ? [command] : (commands || []);
    
        if (safe_mode) {
          for (const cmd of allCommands) {
            if (isCommandDangerous(cmd)) {
              throw new Error(`命令 "${cmd}" 被安全模式阻止。`);
            }
          }
        }
    
        if (mode === 'sync') {
          const results = [];
          for (const cmd of allCommands) {
            const { stdout, stderr, code } = await executeCommand(cmd, options);
            results.push({
              type: "text",
              text: JSON.stringify({ command: cmd, stdout, stderr, code }, null, 2)
            });
          }
          return { content: results };
        } else { // async mode
          for (const cmd of allCommands) {
            executeCommand(cmd, options).catch(error => {
              // 异步错误不会直接返回给用户,但会被框架的日志系统捕获
              console.error(`异步命令执行失败: ${cmd}`, error);
            });
          }
          return {
            content: [{
              type: "text",
              text: `已为 ${allCommands.length} 个命令启动异步执行。请检查系统日志以了解完成状态。`
            }]
          };
        }
      } catch (error) {
        return {
          content: [
            {
              type: "text",
              text: `Error: ${error instanceof Error ? error.message : String(error)}`
            }
          ],
          isError: true
        };
      }
    }
  • JSON schema defining the input parameters for the cli_tool, including command(s), execution mode (sync/async), timeout, working directory, platform, and safe_mode.
    export const schema = {
      name: "cli_tool",
      description: "执行CLI命令,支持同步/异步模式、超时和安全过滤。",
      type: "object",
      properties: {
        command: { type: "string", description: "要执行的单行命令" },
        commands: {
          type: "array",
          items: { type: "string" },
          description: "要执行的多行命令序列 (与 'command' 互斥)"
        },
        mode: {
          type: "string",
          enum: ["sync", "async"],
          default: "sync",
          description: "执行模式: sync (同步阻塞), async (异步非阻塞)"
        },
        timeout: {
          type: "number",
          minimum: 1,
          default: 60,
          description: "命令执行的超时时间(秒)"
        },
        cwd: {
          type: "string",
          description: "命令执行的工作目录 (绝对路径)"
        },
        platform: {
          type: "string",
          enum: ["auto", "win32", "linux", "darwin"],
          default: "auto",
          description: "强制指定执行命令的操作系统环境"
        },
        safe_mode: {
          type: "boolean",
          default: true,
          description: "是否启用危险命令过滤"
        }
      },
      required: []
    };
  • Dynamic registration of all tools (including 'cli_tool') by scanning src/tools directory, dynamically importing each tool module, extracting handler (default export), schema, and destroy function, and populating global tools array and handlers map. Called on server startup.
    export async function loadTools(reload: boolean = false): Promise<{ [key: string]: (request: ToolRequest) => Promise<ToolResponse> }> {
      // 如果是初始加载且已加载,则直接返回
      if (!reload && isLoaded) return;
    
      // 如果是重新加载,则重置状态
      if (reload) {
        for (const tool of tools) {
          await tool?.destroy?.();
          delete handlers[tool.name];
        }
        tools.length = 0;
        isLoaded = false;
      }
    
      // 获取所有工具文件
      const toolFiles = fs.readdirSync(toolsDir).filter(file => file.endsWith('.js') || file.endsWith('.ts'));
    
      // 加载每个工具
      for (const file of toolFiles) {
        const toolPath = path.join(toolsDir, file);
        try {
          // 如果是重新加载,清除模块缓存
          if (reload) clearModuleCache(toolPath);
    
          // 导入模块,重新加载时添加时间戳防止缓存
          const importPath = 'file://' + toolPath + (reload ? `?update=${Date.now()}` : '');
          const { default: tool, schema, destroy } = await import(importPath);
          const toolName = path.parse(toolPath).name;
    
          // 注册工具
          tools.push({
            name: toolName,
            description: tool.description,
            inputSchema: schema,
            destroy: destroy
          });
    
          // 注册处理函数
          handlers[toolName] = async (request: ToolRequest) => { return await tool(request); };
        } catch (error) {
          console.error(`Failed to ${reload ? 'reload' : 'load'} tool ${file}:`, error);
        }
      }
    
      isLoaded = true;
      if (reload) console.log(`Successfully reloaded ${tools.length} tools`);
    
      return handlers;
    }
  • Core helper function that spawns a child process to execute a single CLI command, captures stdout/stderr, handles close/error events, supports timeout and platform-specific shell.
    async function executeCommand(command: string, options: any): Promise<{ stdout: string; stderr: string; code: number | null }> {
      return new Promise((resolve, reject) => {
        const shell = options.platform === 'win32' ? 'cmd.exe' : '/bin/bash';
        const shellArg = options.platform === 'win32' ? '/c' : '-c';
    
        const spawnOptions: SpawnOptions = {
          cwd: options.cwd,
          env: process.env,
          shell: true,
          timeout: options.timeout * 1000 // 将秒转换为毫秒
        };
    
        const child = spawn(shell, [shellArg, command], spawnOptions);
    
        let stdout = '';
        let stderr = '';
    
        child.stdout?.on('data', (data) => {
          stdout += data.toString();
        });
    
        child.stderr?.on('data', (data) => {
          stderr += data.toString();
        });
    
        child.on('close', (code) => {
          resolve({ stdout, stderr, code });
        });
    
        child.on('error', (err) => {
          if ((err as any).code === 'ETIMEDOUT') {
            reject(new Error(`命令执行超时 (超过 ${options.timeout} 秒)`));
          } else {
            reject(err);
          }
        });
      });
    }
  • Cleanup function called during tool reload or shutdown to log destruction of the cli_tool.
    export async function destroy() {
      console.log("Destroy cli_tool tool");
    }
Behavior1/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Tool has no description.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness1/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Tool has no description.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness1/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Tool has no description.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters1/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Tool has no description.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose1/5

Does the description clearly state what the tool does and how it differs from similar tools?

Tool has no description.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines1/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Tool has no description.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Install Server

Other Tools

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/xiaoguomeiyitian/ToolBox'

If you have feedback or need assistance with the MCP directory API, please join our Discord server