Testing Plugins

This guide covers testing patterns for elizaOS plugins — from unit testing individual actions and providers to integration testing with the runtime, and embedding test suites in your plugin.

Setup

Plugins use Vitest as the test runner. Add it to your plugin's dev dependencies:

json

{
  "devDependencies": {
    "vitest": "^4.0.0"
  },
  "scripts": {
    "test": "vitest run",
    "test:watch": "vitest watch",
    "test:coverage": "vitest run --coverage"
  }
}

Mock Runtime Factory

Most plugin tests need a mock IAgentRuntime. Create a shared helper:

typescript

// tests/helpers.ts
import { vi } from 'vitest';
import {
  AgentRuntime,
  createCharacter,
  createMessageMemory,
  InMemoryDatabaseAdapter,
  stringToUuid,
  type IAgentRuntime,
  type Memory,
  type State,
} from '@elizaos/core';

export function createMockRuntime(
  overrides?: Partial<IAgentRuntime>
): IAgentRuntime {
  const runtime = new AgentRuntime({
    agentId: stringToUuid('test-agent'),
    character: createCharacter({ name: 'Test Agent' }),
    adapter: new InMemoryDatabaseAdapter(),
    plugins: [],
    logLevel: 'error',
  });

  runtime.getSetting = vi.fn((key: string) => process.env[key]);
  runtime.getService = vi.fn();
  runtime.composeState = vi
    .fn()
    .mockResolvedValue({ values: {}, data: {}, text: '' } satisfies State);

  Object.assign(runtime, overrides);
  return runtime;
}

export function createMockMessage(
  text: string,
  overrides?: Partial<Memory>
): Memory {
  return {
    ...createMessageMemory({
      id: stringToUuid('test-message'),
      entityId: stringToUuid('test-user'),
      roomId: stringToUuid('test-room'),
      content: { text },
    }),
    ...overrides,
  };
}

Unit Testing Actions

Test the validate and handler methods independently:

typescript

// tests/actions/weather.test.ts
import { describe, it, expect, vi, beforeEach, afterEach } from 'vitest';
import { checkWeatherAction } from '../../src/actions/weather';
import { createMockRuntime, createMockMessage } from '../helpers';

describe('checkWeatherAction', () => {
  let runtime: ReturnType<typeof createMockRuntime>;

  beforeEach(() => {
    runtime = createMockRuntime({
      getSetting: vi.fn((key) => {
        if (key === 'WEATHER_API_KEY') return 'test-key-123';
        return undefined;
      }),
    });
  });

  describe('validate', () => {
    it('returns true when API key is configured', async () => {
      const message = createMockMessage('What is the weather?');
      const result = await checkWeatherAction.validate(runtime, message);
      expect(result).toBe(true);
    });

    it('returns false when API key is missing', async () => {
      const noKeyRuntime = createMockRuntime();
      const message = createMockMessage('What is the weather?');
      const result = await checkWeatherAction.validate(noKeyRuntime, message);
      expect(result).toBe(false);
    });
  });

  describe('handler', () => {
    it('returns weather data on success', async () => {
      // Mock the fetch call
      globalThis.fetch = vi.fn().mockResolvedValue({
        ok: true,
        json: () => Promise.resolve({ temp: 22, condition: 'Sunny' }),
      });

      const message = createMockMessage('Weather in Tokyo');
      const state: State = { values: {}, data: {}, text: '' };
      const result = await checkWeatherAction.handler(
        runtime,
        message,
        state,
        { parameters: { city: 'Tokyo' } }
      );

      expect(result.success).toBe(true);
      expect(result.text).toContain('Tokyo');
    });

    it('returns error on API failure', async () => {
      globalThis.fetch = vi.fn().mockRejectedValue(new Error('Network error'));

      const message = createMockMessage('Weather in Tokyo');
      const state: State = { values: {}, data: {}, text: '' };
      const result = await checkWeatherAction.handler(
        runtime,
        message,
        state,
        { parameters: { city: 'Tokyo' } }
      );

      expect(result.success).toBe(false);
      expect(result.error).toContain('Network error');
    });
  });

  afterEach(() => {
    vi.restoreAllMocks();
  });
});

Unit Testing Providers

Providers return context strings. Test that the output is well-formatted and contains expected data:

typescript

// tests/providers/status.test.ts
import { describe, it, expect } from 'vitest';
import { pluginStatusProvider } from '../../src/providers/status';
import { createMockRuntime, createMockMessage } from '../helpers';

describe('pluginStatusProvider', () => {
  it('returns active status when API key is set', async () => {
    process.env.WEATHER_API_KEY = 'test-key';
    const runtime = createMockRuntime();
    const message = createMockMessage('hello');

    const state: State = { values: {}, data: {}, text: '' };
    const result = await pluginStatusProvider.get(runtime, message, state);

    expect(result).toBeDefined();
    expect(typeof result.text).toBe('string');
    expect(result.text).toContain('active');

    delete process.env.WEATHER_API_KEY;
  });

  it('returns inactive status when API key is missing', async () => {
    delete process.env.WEATHER_API_KEY;
    const runtime = createMockRuntime();
    const message = createMockMessage('hello');

    const state: State = { values: {}, data: {}, text: '' };
    const result = await pluginStatusProvider.get(runtime, message, state);

    expect(result.text).toContain('missing');
  });
});

Unit Testing Services

Test that services start and stop cleanly without leaking resources:

typescript

// tests/services/weather-cache.test.ts
import { describe, it, expect, vi, afterEach } from 'vitest';
import { WeatherCacheService } from '../../src/services/weather-cache';
import { createMockRuntime } from '../helpers';

describe('WeatherCacheService', () => {
  let service: { stop: () => Promise<void> } | undefined;

  it('starts without errors', async () => {
    const runtime = createMockRuntime();
    service = await WeatherCacheService.start(runtime);
    expect(service).toBeDefined();
    expect(service.stop).toBeTypeOf('function');
  });

  it('stops cleanly', async () => {
    const runtime = createMockRuntime();
    service = await WeatherCacheService.start(runtime);
    await expect(service.stop()).resolves.toBeUndefined();
    service = undefined;
  });

  afterEach(async () => {
    if (service) await service.stop();
  });
});

Integration Testing

For tests that need the full runtime (database, memory, state composition), bootstrap a test runtime:

typescript

// tests/integration/plugin.test.ts
import { describe, it, expect, beforeAll, afterAll } from 'vitest';
import type { IAgentRuntime } from '@elizaos/core';
import weatherPlugin from '../../src/index';

describe('weather plugin integration', () => {
  let runtime: IAgentRuntime;

  beforeAll(async () => {
    // If your test setup bootstraps a real runtime:
    // runtime = await createTestRuntime({ plugins: [weatherPlugin] });

    // Or use a mock with real state composition:
    const runtimeMock = {
      agentId: 'test-agent',
      getSetting: (key: string) => process.env[key],
      logger: console,
      // Add other methods your plugin needs
    } satisfies Partial<IAgentRuntime>;
    runtime = runtimeMock as IAgentRuntime;

    // Initialize the plugin
    if (weatherPlugin.init) {
      await weatherPlugin.init({}, runtime);
    }
  });

  it('registers all actions', () => {
    expect(weatherPlugin.actions).toHaveLength(1);
    expect(weatherPlugin.actions![0].name).toBe('CHECK_WEATHER');
  });

  it('registers all providers', () => {
    expect(weatherPlugin.providers).toHaveLength(1);
    expect(weatherPlugin.providers![0].name).toBe('weatherPluginStatus');
  });

  it('plugin init logs correctly', () => {
    // Verify init was called without errors
    expect(weatherPlugin.name).toBe('weather-plugin');
  });
});

Mocking Patterns

Mocking LLM Responses

When testing actions that call the LLM via runtime.useModel:

typescript

const runtime = createMockRuntime({
  useModel: vi.fn().mockResolvedValue({
    text: 'The weather in Tokyo is 22°C and sunny.',
  }),
});

Mocking Database Calls

typescript

const runtime = createMockRuntime({
  getMemoryManager: vi.fn().mockReturnValue({
    searchMemories: vi.fn().mockResolvedValue([]),
    createMemory: vi.fn().mockResolvedValue(undefined),
  }),
});

Mocking External APIs

Use vi.fn() on globalThis.fetch or inject a mock HTTP client:

typescript

globalThis.fetch = vi.fn()
  .mockResolvedValueOnce({
    ok: true,
    json: () => Promise.resolve({ data: 'first call' }),
  })
  .mockResolvedValueOnce({
    ok: false,
    status: 429,
    statusText: 'Too Many Requests',
  });

TestSuite: Embedded Plugin Tests

Plugins can embed tests via the tests field. These run when users execute eliza plugins test <name>:

typescript

import type { Plugin, TestSuite, Memory, State } from '@elizaos/core';
import { checkWeatherAction } from './actions/weather';
import { pluginStatusProvider } from './providers/status';

const weatherTests: TestSuite = {
  name: 'weather-plugin-tests',
  tests: [
    {
      name: 'action validates with API key',
      fn: async (runtime) => {
        const msg = { content: { text: 'weather' } } as Memory;
        const valid = await checkWeatherAction.validate(runtime, msg);
        if (!valid) throw new Error('Expected validation to pass');
      },
    },
    {
      name: 'provider returns context',
      fn: async (runtime) => {
        const msg = { content: { text: 'status' } } as Memory;
        const state: State = { values: {}, data: {}, text: '' };
        const result = await pluginStatusProvider.get(runtime, msg, state);
        if (!result.text) throw new Error('Expected non-empty text');
      },
    },
  ],
};

const weatherPlugin: Plugin = {
  name: 'weather-plugin',
  description: 'Weather information plugin',
  actions: [checkWeatherAction],
  providers: [pluginStatusProvider],
  tests: [weatherTests],
};

export default weatherPlugin;

Running Tests

bash

# Run all tests
vitest run

# Run with coverage report
vitest run --coverage

# Run a specific test file
vitest run tests/actions/weather.test.ts

# Watch mode (re-runs on file changes)
vitest watch

Coverage Thresholds

The monorepo enforces minimum coverage in vitest.config.ts:

Metric	Minimum
Lines	25%
Functions	25%
Statements	25%
Branches	15%

For standalone published plugins, aim for 80% coverage — this is the recommended bar for quality.

E2E Testing

For end-to-end testing with a running agent, the starter template includes a Cypress scaffold:

my-plugin/
├── cypress/
│   ├── e2e/
│   │   └── plugin.cy.ts
│   └── support/
│       └── commands.ts
└── cypress.config.ts

E2E tests start the agent, load the plugin, and verify behavior through the chat or API:

typescript

// cypress/e2e/plugin.cy.ts
describe('Weather Plugin E2E', () => {
  it('responds to weather queries', () => {
    cy.request('POST', 'http://localhost:18789/api/conversations', {
      title: 'Weather Plugin Test',
    }).then(({ body }) => {
      const conversationId = body.conversation.id;
      cy.request(
        'POST',
        `http://localhost:18789/api/conversations/${conversationId}/messages`,
        { text: 'What is the weather in London?' },
      ).then((response) => {
        expect(response.status).to.eq(200);
        expect(response.body.text).to.include('London');
      });
    });
  });
});

Create a Plugin — Build a plugin from scratch
Plugin Patterns — Common implementation patterns
Plugin Schemas — Full type reference
Contributing Guide — Test conventions for the monorepo

Setup

Mock Runtime Factory

Unit Testing Actions

Unit Testing Providers

Unit Testing Services

Integration Testing

Mocking Patterns

Mocking LLM Responses

Mocking Database Calls

Mocking External APIs

TestSuite: Embedded Plugin Tests

Running Tests

Coverage Thresholds

E2E Testing

Related