Legacy codebase load-bearing weirdness
You've inherited a 15-year-old Java monolith with no docs, sparse tests, and a lot of 'load-bearing' weirdness — code that looks wrong but is compensating for something elsewhere. You want to use a coding agent to start modernizing it, but the agent has no way to know which oddities are bugs and which are deliberate. How do you manage an agent productively on an unfamiliar legacy codebase like this without it 'cleaning up' something that quietly holds the system together?
find_behavior_changes(baseline_lines: list[str], candidate_lines: list[str]) → list[str][["rounds_half_up=2.5","empty_input=ERROR","legacy_pad= x "],["rounds_half_up=2.5","empty_input=ERROR","legacy_pad=x"]]out["legacy_pad"][["a=1","b=2"],["a=1","b=2","c=3"]]out[]Treat the AI’s output as a draft to verify, not an answer to trust. Name the specific flaw and the input that triggers it, say how you’d catch it (tests, edge cases, reading critically), and how you’d re-prompt or decompose to get it right.
Vibe & agentic: describe the solution in plain language (or narrate it) and the coach grades your approach.