What problem does it solve? Unit tests and static checks cannot confirm that a feature actually works end to end across the browser, HTTP API, and database. This Skill runs scenario-based full-chain verification so you can validate real runtime behavior before a PR or release, with controlled test data and best-effort rollback. ## Core Features & Use Cases - Scenario-table driven execution: Takes a scenario-path-expectation table and executes each row using a browser driver, HTTP client, and database client together, logging pass/fail per scenario. - Cross-layer assertions: Verifies UI toasts and state, API response codes, and database field changes in one flow, with MutationObserver-based toast capture and header templates copied from real browser requests. - Playbook accumulation: Reusable environment constants (URLs, auth headers, test data fingerprints, irreversible side effects) are persisted to .trellis/spec/guides/e2e-playbook.md after user confirmation, so future runs skip re-discovery. - Use Case: Before merging a dispatch-approval feature, you provide a scenario table; the Skill drives the UI, calls the API with real auth headers, asserts DB state transitions and audit log entries, then restores the test data and reports coverage. ## Quick Start Ask the AI to run full-chain end-to-end verification for the current task using a scenario table of paths and expected outcomes.
