Use when the user wants to understand how a Rust or Python program actually executed — onboarding into an unfamiliar flow, verifying AI-written code, answering "why did this run do X / take this branch / return this value", or understanding a failing test. Cairn records a REAL run and serves recorded ground truth at three zoom levels. Trigger on: "trace this", "why did <function> return/branch/do X", "walk me through what actually happened", "what was <variable> at runtime", "help me understand this flow/handler/test", "why did this test fail". Do NOT use for static questions answerable by reading source alone, or for languages other than Rust and Python (v0).
2026-06-16