Skip to main content

Testing Agents

test-coverage​

Category: testing

Writes tests for all uncovered functions, targeting 80% line coverage.

Process:

  1. Runs the existing test suite and collects coverage data
  2. Identifies all functions and methods with less than 80% coverage
  3. For each uncovered function, retrieves context and writes targeted tests
  4. Re-runs coverage to verify the 80% target is met

Tests are written in the project's existing test framework (pytest, Jest, etc.) and follow the existing test file naming conventions.

nimbus run --agent test-coverage

integration-test-builder​

Category: testing

Generates integration tests for all API endpoints.

For each endpoint it finds:

  • Creates a test that hits the actual endpoint (not mocked)
  • Tests the happy path (valid input, expected response)
  • Tests common error cases (missing fields, invalid auth, etc.)
  • Verifies response schema and status codes

Works with FastAPI, Flask, Django, Express, and Hono.

nimbus run --agent integration-test-builder

mutation-tester​

Category: testing

Finds weak tests using mutation testing, then rewrites the tests that don't catch mutations.

Process:

  1. Introduces small mutations into the source code (flip a boolean, change an operator, remove a return value)
  2. Runs the test suite against each mutation
  3. Identifies mutations that weren't caught (survived mutations = weak tests)
  4. Rewrites the corresponding tests to assert the specific behavior that was missing

This catches tests that pass but don't actually verify correctness — the most dangerous kind.

nimbus run --agent mutation-tester