← ClaudeAtlas

lilia-app-validationlisted

Validation strategy for LiliaCode native desktop changes. Use when choosing or reporting checks after app code, contracts, UI, platform, build config, documentation, or dependency pins change.
sena-nana/LiliaCode · ★ 7 · AI & Automation · score 69
Install: claude install-skill sena-nana/LiliaCode
# Lilia App Validation Run the smallest check that proves the changed behavior. Prefer targeted functional checks over broad or brittle assertions. Repository command authority is `Agents.md` and `docs/guide/development.md`. - Docs, comments, or ignore-rule cleanup: no desktop system gate. If `.gitignore` or boundary rules changed, run `cargo xtask boundary-check`. - Default repo gate: `cargo xtask verify` (boundary, pin, workspace test, workspace check). - Desktop-only Rust: `cargo check --locked -p lilia-desktop` and `cargo test --locked -p lilia-desktop`. - UI main paths, Agent runtime, persistence, permission, contracts, or user-critical flows: `cargo xtask agent-debug`. See `$lilia-agent-debug`. - Performance: `cargo xtask performance`. - Android logic: `cargo xtask android test`. Real device: `cargo xtask android smoke`. - Windows release packaging: `cargo xtask release windows --tag <tag>` then `cargo xtask installer-smoke --tag <tag>`. ## Test Quality - Add tests only for behavior changes or meaningful regression risk. - Do not add tests for documentation-only, comment-only, or formatting-only changes. - Do not write tests that only hard-match log text, incidental strings, or snapshot-like markup. - Assert user-visible behavior, command results, persisted records, permission outcomes, or data-contract handling. ## Reporting - State exactly which checks ran and whether they passed. - If a check was skipped, explain why it was not necessary. - If a check cannot ru