name: snagbadger-test description: Run a SnagBadger catalog or selected scenarios with the configured agent/browser runtime and display retained test results and evidence.#

Find the SnagBadger runner checkout, target catalog/runtime, and onboarding readiness record. Use existing onboarding discovery for application candidates. Ask which application when several are available, even with a default; include all configured catalogs for that application unless explicitly narrowed. Resolve the requested selection; explain unmet prerequisites before execution. An onboarding offer to run tests is not itself acceptance of that offer. API/CLI tests remain independent of missing browser capabilities. Use explicit ready IDs when selected scenarios have blocked prerequisites, and tell the user which tests were excluded and why.

Delegate coordination to the installed run-test-plan skill, or read its copy at <runner-root>/skills/execution/run-test-plan/SKILL.md when not installed. That maintained skill owns bridge requests, fresh workers/contexts, hooks, evidence, retries, cancellation, and HTML reports. Read the runner's current docs/integration/agent-testing.md and runtime mappings; do not create a parallel execution engine in these instructions.

Preserve failed, blocked, incomplete, and flaky observations and all attempts. Display the actual generated report and summarize executed/excluded selections, outcomes, and capability gaps. Use create-test-report/create-test-bug through the maintained runner when needed. Live tracker submission, automatic fixes, and changes to expected behavior require their own authorization.