What problem does it solve? Config-shaped deliverables like lifecycle hooks, injected prompt blocks, status indicators, and settings keys have no test suite to run, so a change that parses and registers can still silently never fire. This Skill closes that gap by mapping each surface class to its cheapest runtime probe, proving the mechanism fired and the intended behavior changed before any completion claim. ## Core Features & Use Cases - Wired/Fired/Effected Ladder: Enforces three verification rungs so a config that merely parses is never mistaken for one that actually ran. - Probe Table by Surface Class: Maps hooks, event wiring, settings keys, injected context blocks, status indicators, preprocessors, command definitions, agent definitions, auto-triggers, git hooks, and scheduled jobs to concrete fire/effect probes. - Dead-Event Diagnosis: Cross-checks a co-resident always-on handler on the same event to distinguish a dead runtime event from a broken matcher. - Use Case: After adding a completion hook to your agent harness, run the hook script directly with a synthetic event payload and paste its stdout as evidence, then trigger the real event once in a scratch context and confirm the marker file updated. ## Quick Start Ask the agent to verify that your newly configured hook or injected context block actually fires at runtime and to paste the probe evidence before marking the task done.
