Improve Coverage
Testing & Qualityimprove-coverageImprove Coverage is an Agent Skill for testing & quality published in the redpanda-data/redpanda repository on GitHub: a SKILL.md instruction file an AI agent can load to do this task.
Improve code coverage for a Bazel C++ test target or source file
How to use this skill
Bring this guide into your coding agent with a prompt tailored to the tool you use.
- Open your project in Codex.
- Copy and paste the prompt below.
- Review files and risks before approving any action.
Inspect the Agent Skill at https://github.com/redpanda-data/redpanda/blob/HEAD/.claude/skills/improve-coverage/SKILL.md and all supporting files. Report provenance, permissions, scripts, dependencies, and security risks. Propose an exact installation into .agents/skills/improve-coverage, but wait for my approval before writing files or running scripts.
Copying this prompt does not install or run the skill. Review third-party files before use.
Improve Code Coverage
Improve test coverage for a Bazel C++ test target or source file path.
Input Resolution
-
Bazel test target (starts with
//, e.g.//src/v/bytes/tests:iobuf_test): use it directly. -
Source file path (e.g.
src/v/bytes/iobuf.cc): find the test target:bazel query "kind('.*_test', rdeps(//src/v/..., //$(dirname $FILE):$(basename $FILE .cc), 1))" 2>/dev/nullIf multiple targets match, pick the one in the closest
tests/directory. If none match, tell the user and stop.
Workflow
Step 1: Baseline Coverage
Run coverage and read the LLM report:
tools/run-cov <target> -r llm
Note the total line coverage percentage — this is the baseline to beat.
Read the report and identify:
- Uncovered functions — highest impact. One test can cover many lines.
- Partially covered functions — uncovered branches indicate missing error-path or edge-case tests.
- Zero-coverage files in scope — may need new test cases entirely.
Step 2: Understand the Code
Read the source files named in the report. Focus on:
- Uncovered functions: what they do, inputs, preconditions.
- The existing test file: patterns, fixtures, helpers.
Key codebase patterns:
redpanda_cc_gtesttargets use GTest (TEST(),TEST_F()).redpanda_cc_btesttargets use Seastar Boost test (SEASTAR_THREAD_TEST_CASE). Match whichever the test file uses.- Many functions are
ss::future<>coroutines — test withSEASTAR_THREAD_TEST_CASEor the Seastar GTest runner. - Use
ssnamespace prefix for Seastar types. - Use
EXPECT_*for most checks,ASSERT_*only when continuing would crash.
Step 3: Write Tests
Prioritize by impact:
- Uncovered functions — call each one. A 20-line function = 20 lines covered.
- Uncovered error paths — trigger the uncovered branches in partially covered functions.
- Zero-coverage files — add smoke tests if practical.
Rules:
- Add tests to the existing test file. Don't create new files or BUILD targets unless necessary.
- Follow existing naming conventions in the file.
- One logical behavior per test case.
- Prefer deterministic tests — construct error conditions directly.
- Do NOT add comments that restate the code.
Step 4: Verify Build
bazel build <target>
Fix compilation errors before proceeding.
Step 5: Verify Coverage
Re-run coverage:
tools/run-cov <target> -r llm
Compare to the baseline. Report:
Coverage: X.Y% → A.B% (+N.N%)
New lines covered: <count>
If coverage did not improve, investigate:
- Build errors preventing the new tests from running?
- Code compiled out by preprocessor guards?
- Template instantiations not triggered by test types?
Step 6: Summary
Report:
- Which functions/paths are now covered
- What test cases were added
- Before/after coverage numbers
- Remaining significant gaps and suggestions