The summary aggregation labels (Count, Sum, Avg, Min, Max) are hardcoded in English in the formatSummaryLabel function. For a global product, these labels should be internationalized using the i18n system. Consider using a translation key instead of hardcoded strings, similar to how other UI text in the codebase is handled. Reference the CRM example's i18n implementation at examples/crm/src/i18n/ for the pattern.

There's a logic issue in the computeAggregation function on line 41-42. When computing a 'count' aggregation, it returns values.length, which is the count of numeric values after filtering (lines 100-109). However, for 'count' aggregations, the correct behavior is handled separately on lines 113-115 which counts all non-null values regardless of type. The 'count' case in this function should never be reached because of the special handling on line 113, but having it here with incorrect logic could cause confusion. Consider removing the 'count' case from the switch statement since it's handled separately, or add a comment explaining why it's unreachable code.

The PR title contains Chinese text "列级高级特性实现" which violates the English-only codebase requirement. All user-facing text, titles, documentation, and comments must be written in English to ensure global accessibility and consistency. Please update the PR title to use only English text, for example: "feat: Column pinned/summary/link/action — Column-Level Advanced Features Implementation"

The summary footer is only rendered when the grid is not grouped (line 1270-1275). When grouping is enabled, the summary aggregations are not displayed at all. This creates an inconsistent user experience where summary configurations on columns are silently ignored when grouping is active. Consider either: (1) rendering the summary footer after all groups in the grouped view, or (2) documenting this limitation in the PR description and adding a comment in the code explaining why summaries are not shown in grouped mode.

The implementation passes the full data array to useColumnSummary on line 364, which means summaries are computed over all data rows, not just the currently visible/filtered rows. If the grid has pagination or filtering enabled, the summary values will include data from all pages/hidden rows, which may be misleading to users who expect to see summaries only for the visible data. Consider passing the filtered/paginated data subset instead, or document this behavior clearly in the component documentation.

Test coverage is missing for the interaction between summary features and grouped views. Since the implementation only renders summaries in non-grouped mode (line 1270-1275 in ObjectGrid.tsx), there should be tests verifying: (1) that summaries are not displayed when grouping is enabled, or (2) that summaries are correctly aggregated per-group if that's the intended behavior. Add test cases covering the grouped + summary scenario.

The action button on line 522-536 lacks an accessible label for screen readers. While the visible text shows the formatted action label, there's no aria-label or similar attribute to indicate the button's purpose in the context of the row. Consider adding an aria-label that includes context, for example: aria-label={${formatActionLabel(col.action!)} for ${row.name || row._id}} to provide screen reader users with more information about which row the action will be performed on.

The Actions column (added on lines 753-793) is added before pinned column reordering logic executes (lines 795-812). If the Actions column should always appear at the far right, it might conflict with user-specified right-pinned columns. Consider either: (1) explicitly setting pinned: 'right' on the Actions column so it follows the same pinning logic as other columns, or (2) adding the Actions column after the pinned reordering logic so it always appears last. The current implementation may result in unexpected column ordering when users specify right-pinned columns.

Test coverage is missing for summary aggregation behavior with pagination and filtering. The current tests only verify summaries with static data arrays. Add test cases that verify: (1) whether summaries aggregate over all data or only visible/filtered data when pagination is enabled, and (2) how summaries behave when filters are applied. This is important to document the expected behavior given the potential bug identified in line 364 of ObjectGrid.tsx.