palarran-app-verify

原文涵盖的主题

• /palarran:app:verify {target/scenarios} [--plan]

Begin with task context, target instructions and adapter selection. Read the relevant branch below, not every adapter guide. Reuse the task's existing decisions, context and bounded memory results; missing optional memory does not block acceptance work. Freeze R-IDs, actor, Given/When/Then, expected business outcomes and the exact reviewed source/deployment subject before checks. Checkers and expected results are independent of the builder. --plan prepares the protocol from available context; it does not run tests, apps, browser/API/DB tools, installs or model calls. Execution uses only already authorized local test/fixture or target access. Never infer approval for production effects, paid providers, new services or downloads. Server/worker/bot project tests: read project-tests. Map existing reviewed commands to R-IDs, use fake providers when the real dependency is outside scope, and keep fixture/projecttest PASS distinct from business acceptance. Do not invent a browser/HTTP/SQLite shape for a server or mark its entire review incomplete just because those surfaces do not exist. Browser/API/database acceptance: read the relevant methodology, contract, operations, evidence and sour

这里只展示节选，请以完整原文和当前要求为准。