Testing Suite

Relevant source files

The following files were used as context for generating this wiki page:

• tests/test_entrypoint.py
• tests/test_integration.py
• tests/test_memory.py
• tests/test_tool_docs.py
• tests/test_resources.py
• mcp_server_code_execution_mode.py

The Testing Suite for the MCP Server Code Execution Mode bridge is designed to ensure the reliability and security of the sandboxed Python execution environment and its interaction with the Model Context Protocol (MCP). It encompasses unit tests for internal logic, integration tests for end-to-end tool calling, and specialized testing for the persistent memory system and documentation discovery mechanisms.

The suite utilizes unittest and unittest.IsolatedAsyncioTestCase to handle the asynchronous nature of MCP communications. It employs mocking and in-process sandbox simulations to validate core logic without requiring a full container runtime during standard test execution.

Sources: tests/test_entrypoint.py:12-14, tests/test_integration.py:12-15, tests/test_memory.py:10-12

Core Testing Components

The testing architecture is divided into specialized modules that target specific layers of the bridge.

Entrypoint Generation and Execution

Tests in test_entrypoint.py focus on the RootlessContainerSandbox._render_entrypoint method. They simulate the container's internal environment by capturing sys.__stdout__ and providing a mock stdin to verify how the generated Python script handles RPC requests and user code execution.

Integration and Stub Servers

Integration tests verify the bridge's ability to discover, load, and call tools from external MCP servers. A StubIntegrationTests class uses a stub_mcp_server.py to simulate a real stdio-based MCP server, ensuring that the MCPBridge correctly routes calls through its PersistentMCPClient.

Sources: tests/test_integration.py:136-160, tests/test_integration.py:186-218

Memory System Validation

The MemorySystemTests validate the persistent data storage layer. These tests simulate the sandbox's filesystem by creating temporary directories and verifying atomic updates, sanitization of keys, and JSON serialization.

Sources: tests/test_memory.py:17-43, tests/test_memory.py:149-166

Testing Infrastructure Summary

Component	Test File	Primary Focus	Key Classes/Methods
Entrypoint Logic	test_entrypoint.py	Sandbox internal script generation	_run_entrypoint, EntryPointGenerationTests
Tool Integration	test_integration.py	End-to-end MCP server communication	InProcessSandbox, StubIntegrationTests
Memory System	test_memory.py	Atomic persistence and sanitization	MemorySystemTests, _save_memory, _update_memory
Documentation	test_tool_docs.py	Discovery and search RPC handlers	ToolDocsTests, search_tool_docs
MCP Resources	test_resources.py	Capability document exposure	ResourceEndpointTests, read_resource

Sources: tests/test_entrypoint.py, tests/test_integration.py, tests/test_memory.py, tests/test_tool_docs.py, tests/test_resources.py

Sandbox Simulation Architecture

To allow testing without a full container runtime (Podman/Docker), the suite uses an InProcessSandbox. This class implements the SandboxLike protocol but executes code locally using exec() while redirecting I/O.

Specific Feature Testing

Documentation Discovery

Tests ensure that the mcp.runtime helpers provided to the LLM are accurate. This includes verifying that fuzzy search and summary queries return the expected tool metadata.

• Fuzzy Search: Validated in test_tool_docs.py to ensure keyword matching across server and tool descriptions.
• Detail Levels: Tests confirm that detail='full' returns the inputSchema while detail='summary' omits it.

Sources: tests/test_tool_docs.py:44-52, tests/test_entrypoint.py:190-210

Memory Persistence and Sanitization

The memory tests specifically verify that the system handles edge cases like special characters in keys and atomic updates to JSON objects.

def test_key_sanitization(self):
    """Test that special characters in keys are sanitized."""
    self._save_memory("key with spaces!", "value1")
    self._save_memory("key/with/slashes", "value2")
    # Should be able to load using original keys
    self.assertEqual(self._load_memory("key with spaces!"), "value1")

Sources: tests/test_memory.py:175-184

Resource Endpoints

The ResourceEndpointTests verify that the bridge correctly advertises its capabilities via standard MCP resources, ensuring the CAPABILITY_RESOURCE_URI returns the SANDBOX_HELPERS_SUMMARY.

Sources: tests/test_resources.py:15-30

Conclusion

The Testing Suite provides a comprehensive validation layer for the MCP Bridge, ensuring that security constraints, tool proxying, and data persistence remain stable. By utilizing in-process simulations and mock stdio streams, the suite maintains high coverage and fast execution without the overhead of container management during development.

Sources: tests/test_integration.py, tests/test_entrypoint.py