"Test Coverage Copilot team on IBM Bob 2.0 hackathon Hackathon"

Team Idea

An agentic system that takes any repo and a design/requirements doc, then automatically finds gaps between what's supposed to work and what's actually tested — not just missing code coverage, but missing behavioral coverage. Subagents in parallel — spin up one subagent per module/directory to independently analyze test coverage, so a multi-file repo gets scanned fast instead of sequentially Document understanding — feed it a PRD or spec doc (PDF/markdown); Bob cross-references stated requirements against existing tests to flag "this requirement has no test", not just "this function has no test" Agent mode — Bob doesn't just suggest tests, it writes them, runs them, and iterates if they fail Parallel tasks — generate tests for multiple flagged gaps simultaneously rather than one at a time

Team Members

Rathod Harsh

Harshr

Submission

Test Coverage Copilot

An IBM Bob agentic system that scans real codebases for untested requirements, generates tests via an Actor skill, and independently validates them via a Critic skill — no shared reasoning between them. Found a real unhandled crash bug in the process.