pattern-exploratory-testinglisted

# pattern-exploratory-testing The catalogue for **exploratory testing** — simultaneous learning, test design, and test execution against a *running* build, steered by risk rather than a pre-written script. This is the investigation that scripted coverage cannot do: `pattern-test-coverage` proves the system does what the ACs *said*; exploration finds what the ACs *didn't say* — the surprising input, the half-finished error path, the two features that collide, the thing that's technically-correct-but-wrong. The two are complements: scripted tests are the safety net, exploration is the search for the holes the net doesn't cover. > **This skill is not part of the `implement-slice` review fan-out.** It does not gate a PR, flip a label, or block a slice. It is a role-neutral capability for an agent that drives a live application — most usefully a **pre-release readiness gate**: poke the assembled release candidate the way a skeptical human would, before it actually lands, and return a ship / no-ship readout with the evidence. It never blocks on a "missing AC" — finding *unknown unknowns* is the job, and a finding here is grounded in a broken **oracle**, not a spec clause. ## When to activate - An agent (or a user) is running a **chartered exploratory session** or a **release-readiness check** against a built, running system — "poke around before we ship", "is this release fit to land", "explore the new feature for surprises", "session-based testing of X". - A bug is suspected b