docs: Update GEMINI.md to strictly enforce make check and 100% coverage

GEMINI.md

@@ -11,23 +11,22 @@ This document provides a set of global instructions and principles for the Gemin
 
 ## Code Quality & Standards
 
-### 1. Linting, Formatting & Testing
-ALWAYS run the full pipeline before completing a task:
-- **Command**: `make check`
-- **Action**: Fix ALL errors, warnings, and coverage failures before verifying the task.
-- **Strictness**: DO NOT run individual linters (e.g., `ruff`, `npm run lint`) in isolation. ALWAYS use `make check` to ensure the entire state is valid.
+### 1. Verification & Quality Gates
+**Strict Rule:** ALWAYS run `make check` before marking a task as complete. This command performs:
+- **Linting & Formatting**: (Ruff, ESLint, Prettier)
+- **Testing**: Runs all unit tests.
+- **Coverage**: Enforces **100% line coverage** for both backend and frontend.
 
-### 2. Testing
-ALWAYS run the full test suite to ensure no regressions:
-- **Backend**: `uv run pytest` (in `/backend`)
-- **Frontend**: `npm run test` (in `/frontend`)
-- **Action**: Every new feature or major refactor MUST include or update relevant tests.
+**Action:**
+- If `make check` fails, the task is **NOT DONE**. Fix ALL errors, warnings, and coverage drops.
+- DO NOT rely on individual commands (like `pytest` or `npm test`) for final verification. `make check` is the single source of truth.
+- **New Code**: Must have 100% test coverage and adhere to best practices.
 
-### 3. Error Handling & Logging
+### 2. Error Handling & Logging
 - **No Print Statements**: Use `logging` in Python and `console.error` (with prefix) in Vue.
 - **Fail Fast**: Let exceptions bubble up to the global handler in `backend/src/main.py`.
 - **Validation**: Use Pydantic models for all API requests and responses.
 
-### 4. Documentation
+### 3. Documentation
 - Refer to `CONTRIBUTING.md` for detailed guidelines.
 - Keep `task.md` and `implementation_plan.md` updated throughout the task.