Improving Code Coverage

Systematic approach to identifying and covering untested code using MCP tools.

Step 1: Identify Coverage Gaps

Use test_coverage to see project-wide coverage gaps:

• Files below threshold with exact uncovered line ranges
• Per-metric values (statements, branches, functions, lines)
• Current thresholds and targets

For a specific file, use file_coverage({ filePath: "src/foo.ts" }) to get per-metric values and uncovered line ranges.

Step 2: Understand Existing Tests

Use test (action: for_file)({ filePath: "src/foo.ts" }) to find which test files already cover the source file. Then use test (action: get)({ fullName: "Suite > test name" }) to understand individual test scope and history.

Step 3: Write Targeted Tests

Focus on the uncovered lines from step 1:

1. Read the source file at the uncovered line ranges
2. Identify untested code paths (branches, error handlers, edge cases)
3. Write tests targeting those specific paths
4. Run with run_tests({ files: ["path/to/test.ts"] })

Step 4: Verify Improvement

1. Use file_coverage to check the file's updated coverage
2. Use test_coverage to verify project-wide improvement
3. Use test_trends to confirm coverage direction is improving

Prioritization

Focus on files with the largest coverage gaps first. Files near the threshold ("at-risk" files) are the highest priority since a small regression could push them below the threshold.

Recording Progress

Use note (action: create) to document:

• Which files you improved and by how much
• Untested paths that were intentionally deferred
• Coverage targets agreed upon with the team