What problem does it solve? Manual QA against acceptance criteria is slow and hard to audit: testers must walk flows in a browser, remember to capture evidence, and write up results by hand. This Skill automates that pass by driving a browser through a scenario, capturing a screenshot before every check, and persisting a self-contained run folder with markdown and JSON reports. ## Core Features & Use Cases - Acceptance-criteria-driven runs: Walks a numbered list of criteria in order, evaluates each as passed or failed, and records notes plus an evidence screenshot per criterion. - Self-contained run artifacts: Creates projects/<project>/qa/runs/<timestamp>-<env>-<ticket>-<label>/ containing run.md, run.json, and a screenshots/ folder. - Per-project configuration: Resolves envs, accounts, saved AC files, flows, and helper overrides from projects/<project>/qa/ with shared fallbacks. - Use Case: Before signing off on ticket ACCM-123, ask the agent to verify its acceptance criteria on the local environment; it logs in, walks the flow, screenshots each check, and produces a pass/fail report you can attach to the ticket. ## Quick Start Run /pavilio-qa-agent with --project=myapp --env=local --ticket=ACCM-123 and paste the start URL plus the numbered acceptance criteria to verify.
