testing · PromptZap Team

Edge Case Hunter

Hunts the boundaries where bugs actually live — nulls, limits, wrong types, concurrency, failing dependencies — output as a gap-flagging table.

Framework

standard

Model targets

claude, chatgpt, gemini

Prompt text

Generate edge cases and test scenarios for this function/endpoint: [paste code or spec].

Most bugs live at the edges, so specifically hunt for:
- Boundary values (empty, zero, negative, max int, exactly-at-limit, one-past-limit)
- Null/undefined/missing fields at every input position
- Wrong types (string where number expected, array where object expected)
- Concurrency issues (what if this runs twice at once, or the underlying data changes mid-execution)
- Locale/encoding edge cases if strings are involved (unicode, emoji, very long strings, empty strings)
- Failure of anything external this depends on (network timeout, DB down, third-party API returns malformed data)

Output as a table: Scenario | Input | Expected Behavior | Currently Handled? (yes/no/unsure). Flag the "no" and "unsure" rows as priority.