Abstract

Developers are increasingly overwhelmed by AI-generated issue reports that lack actionability and reproducibility, eroding trust in automated bug detection tools. In this paper, we present IssueSpecter, an automated tool that finds bugs in uncovered code segments and generates prioritized, actionable issue reports. IssueSpecter combines coverage analysis with LLM-based defect identification, producing structured reports complete with severity ratings, reproduction steps, and suggested fixes. Our evaluation on 13 actively maintained Python projects generates 10,467 issue reports across all uncovered code segments, from which a rule-based ranking orders all issues per project. From this initial rule-based ranking, we filter the top 10 highest-severity, most descriptive issues per project, resulting in 130 candidate reports that are subsequently re-ranked by an LLM according to impact, scope, and urgency. Our manual investigation of these 130 issues confirms that 84.6% are valid or warrant further investigation, with only 15.4% classified as false positives. Furthermore, the LLM-based ranking outperforms rule-based ranking by 50% at P@3 and 41% in MRR. The identified bugs span a wide variety of types, from logic and boundary errors to security vulnerabilities and state consistency bugs. We further validate IssueSpecter through case studies reproducing real bugs surfaced from its generated reports, demonstrating its practical value for automatic bug discovery in real open-source Python projects. Furthermore, IssueSpecter achieves a higher valid bug rate in relation to CoverUp.