Skip to main content
Glama

execute_code

Run Python, Go, or JavaScript code securely in isolated containers with the Isolator MCP Server. Execute custom code snippets or predefined templates in a controlled environment.

Instructions

Executes code (Python, Go, JavaScript) in a secure, isolated container environment.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
additional_filesNoOptional array of additional files needed for execution
entrypoint_codeNoThe main code content to execute. Required unless using snippet_name.
entrypoint_filenameNoOptional filename for the main code (defaults based on language).
languageNoThe programming language (python, go, javascript). Required unless using snippet_name.
snippet_nameNoName of a pre-defined code snippet to execute (e.g., 'hello_world'). Mutually exclusive with entrypoint_code/language.

Implementation Reference

  • The primary handler for the 'execute_code' tool within the CallToolRequestSchema handler. It validates input with Zod, handles snippet loading or direct code, creates temporary files, executes code via the isolator CLI using child_process.spawn, parses JSON output, formats the result as MCP CallToolResult, and cleans up temporary directories.
    server.setRequestHandler(CallToolRequestSchema, async (request): Promise<CallToolResult> => {
      if (request.params.name !== "execute_code") {
        throw new McpError(ErrorCode.MethodNotFound, `Unknown tool: ${request.params.name}`);
      }
    
      // Validate input using Zod schema
      const parseResult = executeCodeInputSchema.safeParse(request.params.arguments);
      if (!parseResult.success) {
        console.error("Invalid input:", parseResult.error.errors);
        throw new McpError(ErrorCode.InvalidParams, `Invalid arguments: ${parseResult.error.errors.map(e => `${e.path.join('.')}: ${e.message}`).join(', ')}`);
      }
    
      let languageInput: string;
      let codeInput: string;
      let languageDetectedFromSnippet = false;
      const { entrypoint_filename, additional_files, snippet_name } = parseResult.data;
      let finalEntrypointFilename: string;
    
      // --- Determine code input and language: Snippet or Direct ---
      if (snippet_name) {
        const __filename = fileURLToPath(import.meta.url);
        const __dirname = path.dirname(__filename);
        const snippetsDir = path.resolve(__dirname, '..', serverConfig.promptsDir); // Use config value
        const snippetExt = path.extname(snippet_name);
        const baseSnippetName = snippetExt ? path.basename(snippet_name, snippetExt) : snippet_name;
    
        let foundSnippetPath: string | undefined;
        let detectedLang: string | undefined;
        let snippetFilename: string | undefined;
    
        const possibleExtensions: { [key: string]: string } = {
            ".py": "python",
            ".go": "go",
            ".js": "javascript",
            ".mjs": "javascript",
            ".cjs": "javascript",
        };
    
        if (snippetExt && possibleExtensions[snippetExt]) {
            detectedLang = possibleExtensions[snippetExt];
            foundSnippetPath = path.join(snippetsDir, `${baseSnippetName}${snippetExt}`);
            snippetFilename = `${baseSnippetName}${snippetExt}`;
        } else {
            for (const [ext, lang] of Object.entries(possibleExtensions)) {
                 const testPath = path.join(snippetsDir, `${baseSnippetName}${ext}`);
                 try {
                     await fs.access(testPath);
                     foundSnippetPath = testPath;
                     detectedLang = lang;
                     snippetFilename = `${baseSnippetName}${ext}`;
                     break;
                 } catch { /* continue */ }
            }
        }
    
        if (!foundSnippetPath || !detectedLang) {
            throw new McpError(ErrorCode.InvalidParams, `Snippet '${snippet_name}' not found in '${snippetsDir}' or language could not be determined.`);
        }
    
        try {
            codeInput = await fs.readFile(foundSnippetPath, 'utf-8');
            languageInput = detectedLang;
            languageDetectedFromSnippet = true;
            console.error(`Loaded code from snippet: ${foundSnippetPath}`);
        } catch (readError) {
            throw new McpError(ErrorCode.InternalError, `Failed to read snippet file '${foundSnippetPath}': ${readError instanceof Error ? readError.message : String(readError)}`);
        }
    
        if (!snippetFilename) {
            throw new McpError(ErrorCode.InternalError, `Could not determine snippet filename for ${snippet_name}`);
        }
        finalEntrypointFilename = snippetFilename;
    
      } else {
        languageInput = parseResult.data.language!;
        codeInput = parseResult.data.entrypoint_code!;
    
        finalEntrypointFilename = entrypoint_filename || "";
        if (!finalEntrypointFilename) {
            switch(languageInput.toLowerCase()) {
                case "python": finalEntrypointFilename = "main.py"; break;
                case "go": finalEntrypointFilename = "main.go"; break;
                case "javascript": finalEntrypointFilename = "index.js"; break;
                default:
                    throw new McpError(ErrorCode.InternalError, `Cannot determine default entrypoint filename for ${languageInput}`);
            }
        }
      }
    
      const langLower = languageInput.toLowerCase();
    
      if (!serverConfig.languageSettings[langLower]) {
          const errorMsg = languageDetectedFromSnippet
            ? `Language '${languageInput}' (detected from snippet '${snippet_name}') is not configured.`
            : `Unsupported language provided directly: ${languageInput}`;
          throw new McpError(ErrorCode.InvalidParams, errorMsg);
      }
    
      // --- Temporary Directory and File Creation ---
      let tempDir: string | undefined;
      try {
        tempDir = await fs.mkdtemp(path.join(os.tmpdir(), "isolator-mcp-"));
        // Set directory permissions to be accessible (rwxr-xr-x)
        await fs.chmod(tempDir, 0o755);
        console.error(`Created temp dir: ${tempDir} with mode 755`);
    
        // Write entrypoint file
        const entrypointPath = path.join(tempDir, finalEntrypointFilename);
        await fs.writeFile(entrypointPath, codeInput);
        // Set file permissions to be readable (rw-r--r--)
        // Use 0o755 if script needs to be executable itself by the container user (less common)
        await fs.chmod(entrypointPath, 0o644);
        console.error(`Wrote entrypoint file: ${entrypointPath} with mode 644`);
    
    
        // Write additional files
        if (additional_files) {
          for (const file of additional_files) {
            const safeFilename = path.basename(file.filename);
            if (safeFilename !== file.filename || safeFilename === "." || safeFilename === "..") {
                 throw new McpError(ErrorCode.InvalidParams, `Invalid additional filename: ${file.filename}`);
            }
            const additionalFilePath = path.join(tempDir, safeFilename);
            await fs.writeFile(additionalFilePath, file.content);
            // Set file permissions to be readable (rw-r--r--)
            await fs.chmod(additionalFilePath, 0o644);
            console.error(`Wrote additional file: ${additionalFilePath} with mode 644`);
          }
        }
    
        // --- Construct Go CLI Command ---
        const args: string[] = [
            "run",
            "--language", langLower,
            "--dir", tempDir,
            "--timeout", `${serverConfig.defaultTimeoutSeconds}s`,
            "--memory", String(serverConfig.defaultMemoryMB),
            "--cpus", String(serverConfig.defaultCpus),
            "--pids-limit", String(serverConfig.defaultPidsLimit),
            "--network", serverConfig.defaultNetworkMode,
            "--read-only", String(serverConfig.defaultReadOnlyRootfs),
            "--container-workdir", serverConfig.defaultContainerWorkdir,
            "--entrypoint", finalEntrypointFilename,
            ...serverConfig.defaultCapDrop.map(cap => ["--cap-drop", cap]).flat(),
        ];
    
        // Add GOCACHE env var specifically for Go
        if (langLower === "go") {
            args.push("--env", "GOCACHE=/tmp");
        }
    
    
        console.error(`Executing: ${serverConfig.isolatorPath} ${args.join(" ")}`);
    
        // --- Execute Go CLI ---
        const executionPromise = new Promise<ExecutionResult>((resolve, reject) => {
            // Use default spawn options, relying on the --env flag now
            const spawnOptions: Parameters<typeof spawn>[2] = { stdio: ["ignore", "pipe", "pipe"], env: process.env };
    
            const child = spawn(serverConfig.isolatorPath, args, spawnOptions);
    
            let stdoutData = "";
            let stderrData = "";
    
            if (child.stdout) {
                child.stdout.on("data", (data) => { stdoutData += data.toString(); });
            } else {
                console.error("Warning: child.stdout is null");
            }
    
            if (child.stderr) {
                child.stderr.on("data", (data) => {
                    stderrData += data.toString();
                    console.error("isolator CLI stderr:", data.toString());
                });
            } else {
                 console.error("Warning: child.stderr is null");
            }
    
            child.on("error", (err) => {
                console.error("Failed to start isolator CLI:", err);
                reject(new Error(`Failed to start isolator process: ${err.message}`));
            });
    
            child.on("close", (code) => {
                console.error(`isolator CLI exited with code ${code}`);
                try {
                    const result: ExecutionResult = JSON.parse(stdoutData);
                    resolve(result);
                } catch (parseError) {
                     console.error("Failed to parse isolator JSON output:", parseError);
                     console.error("isolator stdout:", stdoutData);
                     resolve({
                         status: "error",
                         exitCode: code ?? -1,
                         stdout: stdoutData,
                         stderr: stderrData,
                         durationMs: 0,
                         error: `Failed to parse execution result from isolator CLI. Exit code: ${code}. ${parseError instanceof Error ? parseError.message : String(parseError)}`
                     });
                }
            });
        });
    
        const result = await executionPromise;
    
        // --- Format MCP Result ---
        const outputText = `--- stdout ---\n${result.stdout}\n--- stderr ---\n${result.stderr}`;
        const content: TextContent = { type: "text", text: outputText };
        const callResult: CallToolResult = { content: [content] };
    
        if (result.status !== "success") {
            callResult.isError = true;
            content.text = `Execution Failed (${result.status}): ${result.error || 'Unknown error'}\n\n` + content.text;
        }
    
        return callResult;
    
      } catch (error) {
        console.error("Error during execute_code handling:", error);
        throw new McpError(ErrorCode.InternalError, `Failed to execute code: ${error instanceof Error ? error.message : String(error)}`);
      } finally {
        // --- Cleanup Temporary Directory ---
        if (tempDir) {
          try {
            await fs.rm(tempDir, { recursive: true, force: true });
            console.error(`Cleaned up temp directory: ${tempDir}`);
          } catch (cleanupError) {
            console.error(`Error cleaning up temp directory ${tempDir}:`, cleanupError);
          }
        }
      }
    });
  • Zod schemas for validating the input parameters of the 'execute_code' tool: additionalFileSchema and the main executeCodeInputSchema with refinement for mutual exclusivity between snippet_name and direct code inputs.
    const additionalFileSchema = z.object({
        filename: z.string().describe("Name of the file (including extension)"),
        content: z.string().describe("Content of the file"),
    });
    
    // Input schema allowing either direct code OR a snippet name
    const executeCodeInputSchema = z.object({
      language: z.string().optional().describe("The programming language (python, go, javascript). Required unless using snippet_name."),
      entrypoint_code: z.string().optional().describe("The main code content to execute. Required unless using snippet_name."),
      entrypoint_filename: z.string().optional().describe("Optional filename for the main code (defaults based on language)"),
      additional_files: z.array(additionalFileSchema).optional().describe("Optional array of additional files needed for execution"),
      snippet_name: z.string().optional().describe("Name of a pre-defined code snippet to execute (e.g., 'hello_world'). Mutually exclusive with entrypoint_code/language.")
    }).refine(data => (data.snippet_name && !data.entrypoint_code && !data.language) || (!data.snippet_name && data.entrypoint_code && data.language), {
      message: "Either 'snippet_name' or BOTH 'entrypoint_code' and 'language' must be provided, but not both sets.",
    });
  • src/index.ts:106-155 (registration)
    Registration of the 'execute_code' tool in the ListToolsRequestHandler, providing name, description, and inputSchema for MCP tool discovery.
    server.setRequestHandler(ListToolsRequestSchema, async () => {
      return {
        tools: [
          {
            name: "execute_code",
            description: "Executes code (Python, Go, JavaScript) in a secure, isolated container environment.",
            inputSchema: {
              type: "object",
              properties: {
                language: {
                  type: "string",
                  description: "The programming language (python, go, javascript). Required unless using snippet_name."
                },
                entrypoint_code: {
                  type: "string",
                  description: "The main code content to execute. Required unless using snippet_name."
                },
                entrypoint_filename: {
                  type: "string",
                  description: "Optional filename for the main code (defaults based on language)."
                },
                additional_files: {
                  type: "array",
                  description: "Optional array of additional files needed for execution",
                  items: {
                    type: "object",
                    properties: {
                      filename: {
                        type: "string",
                        description: "Name of the file (including extension)"
                      },
                      content: {
                        type: "string",
                        description: "Content of the file"
                      },
                    },
                    required: ["filename", "content"],
                  },
                },
                snippet_name: {
                  type: "string",
                  description: "Name of a pre-defined code snippet to execute (e.g., 'hello_world'). Mutually exclusive with entrypoint_code/language.",
                },
              },
              // Rely on Zod validation in the handler for mutual exclusivity.
            },
          },
        ],
      };
    });
Behavior2/5

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

No annotations are provided, so the description carries full burden. It mentions 'secure, isolated container environment' which hints at safety, but doesn't disclose critical behaviors like execution time limits, resource constraints, output handling, error behavior, or authentication needs. For a code execution tool with zero annotation coverage, this is insufficient.

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

Conciseness5/5

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

The description is a single, efficient sentence that front-loads the core purpose. Every word earns its place, with no redundancy or unnecessary elaboration. It's appropriately sized for the tool's complexity.

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

Completeness2/5

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

Given the complexity of a code execution tool with no annotations and no output schema, the description is incomplete. It lacks details on return values, error handling, execution constraints, and safety guarantees. The description doesn't compensate for the missing structured data, leaving significant gaps for an AI agent.

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

Parameters3/5

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

Schema description coverage is 100%, so the schema fully documents all 5 parameters. The description doesn't add any parameter-specific details beyond what's in the schema, such as explaining mutual exclusivity rules or language-specific defaults. Baseline 3 is appropriate when schema does the heavy lifting.

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

Purpose4/5

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

The description clearly states the action ('Executes code') and resource ('in a secure, isolated container environment'), specifying supported languages (Python, Go, JavaScript). It distinguishes from potential alternatives by mentioning the execution environment, but without sibling tools, full differentiation isn't possible.

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

Usage Guidelines2/5

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

No guidance on when to use this tool versus alternatives is provided. The description mentions the execution environment but doesn't specify use cases, prerequisites, or limitations. Without siblings, this is less critical, but still a gap in usage context.

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

Related 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/Ompragash/isolator-mcp'

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