Goal

• Given the passed function, file, or files <FILES>, increase unit test coverage from the corresponding files <TEST_FILES> to approach 100 percent using a plan-then-implement approach

Workflow

Establish Current Coverage

• Identify the test files corresponding to the target function, file, files <TEST_FILES> using the rules

  • E.g.,

    helpers/hllm_cli.py -> helpers/test/test_hllm_cli.py
    

• Print the files to test

  Target files: <FILES>
  Test files: <TEST_FILES>
  

• Run the following command and record the results:

  # Using invoke (preferred):
  > i run_coverage --suite fast --generate-html-report
  # Or directly with pytest inside Docker:
  > pytest --cov=<PKG> --cov-report=term-missing --cov-report=html <TEST_FILE>
  

• Summarize which lines and branches are not covered

Design a Testing Plan

• Analyze uncovered code paths and behaviors
• Propose a clear testing plan that:
  • Tests functions modularly before testing interactions
  • Prioritizes end-to-end and user-visible behavior over incidental implementation details
  • Avoids testing assertions, defensive checks, or invalid inputs unless they are critical to correctness
• Explicitly map each uncovered area to one or more proposed tests

Specify Tests Without Implementation

• Define test classes and test method names
• For each proposed test:
  • Describe the input data
  • Describe the expected output or observable behavior
• Do not write executable test code at this stage

Implement After Approval

• Wait for user confirmation or feedback on the testing plan
• After approval:
  • Implement the tests according to the approved plan
  • Re-run coverage and verify that coverage for the target function is near 100 percent
  • Report final coverage results and remaining gaps, if any

Important Rules

• Do not implement tests before the plan is approved
• Do not proceed if instructions or requirements are ambiguous
• Always adhere to the formatting and style guides referenced above
• Focus on meaningful behavior and correctness rather than superficial coverage
• All tests and code must strictly follow:
  • .claude/skills/coding.rules.md for any Python code
  • .claude/skills/testing.rules.md for tests
• If any part of the task is unclear or underspecified, stop and ask for clarification before proceeding

Verification

• Confirm coverage for the target function or file is near 100 percent
• Confirm every proposed test was implemented and passes
• Confirm no uncovered critical branch remains unexplained