What problem does it solve?
Reviewing a Java OpenInference instrumentor package by hand is error-prone: reviewers must check Gradle setup, span lifecycle correctness, exhaustive test assertions, and semantic convention compliance, all while verifying behavior against the actual instrumented library source. This Skill automates that audit and reports findings organized by severity.
Core Features & Use Cases
- Structured four-section review: Checks Gradle setup (compileOnly dependencies, settings.gradle registration, Spotless formatting), testing patterns (exhaustive attribute assertions, error handling, TraceConfig masking), OpenInference semantic conventions, and span lifecycle (span.end(), Scope closing, OITracer usage).
- Ground-truth verification: Locates the instrumented library version from the Gradle ext block and reads its actual source from the Gradle cache or Maven Central before flagging any finding, calibrating severity by real runtime behavior.
- Severity-ranked reporting: Presents findings in a Critical/High/Medium/Low table with file paths and line numbers, plus a list of what is working well.
- Use Case: A contributor opens a PR adding a new instrumentor under java/instrumentation/. Ask the Skill to review the package, and it audits the code against project standards and returns a prioritized findings table.
Quick Start
Review the Java instrumentor package for my library and report any correctness or convention issues with severity levels.
