Testing Tools

Testing and quality assurance tools for AI agents, exposed via the Model Context Protocol (MCP).

Tools

generate_tests

Generate test cases from a function signature. Produces four categories of tests as ready-to-run Jest or Vitest code:

• Happy path -- valid inputs, expected outputs
• Edge cases -- empty strings, zero values, null parameters
• Error cases -- missing arguments, wrong types
• Boundary values -- extreme numbers, large arrays, NaN

Parameters:

Name	Type	Required	Description
signature	string	Yes	Function signature, e.g. async function fetchUser(id: number): Promise<User>
framework	string	No	jest or vitest (default: vitest)
module_path	string	No	Import path for the module under test (default: ./module)

generate_mock_data

Generate realistic mock data for testing. Supported types:

name, email, address, date, uuid, phone, company, credit_card, ip

Parameters:

Name	Type	Required	Description
type	string	Yes	Data type to generate
count	number	No	Number of items (default: 10, max: 1000)
locale	string	No	en or es (default: en)
types	string[]	No	Generate mixed records with multiple field types

generate_api_mock

Generate mock API responses from a schema definition. Creates realistic JSON payloads for REST endpoints by inferring values from field names and types.

Parameters:

Name	Type	Required	Description
endpoint	string	Yes	API endpoint path
method	string	No	HTTP method (default: GET)
fields	object[]	Yes	Field schemas with name, type, optional items, fields, nullable, enum
count	number	No	Number of records (default: 1, max: 100)
status_code	number	No	HTTP status code (default: 200)
wrap_in_envelope	boolean	No	Wrap in { success, data, meta } (default: true)

analyze_test_coverage

Analyze source code and test code to find untested functions. Prioritizes suggestions by:

• Export status (public API surface)
• Cyclomatic complexity estimate
• Parameter count
• Async functions (more error paths)

Parameters:

Name	Type	Required	Description
source_code	string	Yes	Source code to analyze
test_code	string	Yes	Existing test code
source_file_name	string	No	Filename label (default: source.ts)

generate_assertions

Generate detailed assertion code by comparing expected and actual values. Performs deep diff and produces per-field assertions with descriptive comments.

Parameters:

Name	Type	Required	Description
expected	string	Yes	Expected value as JSON string
actual	string	Yes	Actual value as JSON string
label	string	No	Description for the comparison
framework	string	No	jest, vitest, or chai (default: jest)
deep	boolean	No	Deep equality for objects/arrays (default: true)

Setup

npm install
npm run build

Usage with Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "testing-tools": {
      "command": "node",
      "args": ["path/to/mcp-testing-tools/dist/index.js"]
    }
  }
}

License

MIT