memory-structural-dedupe

To fix duplicate sections in a markdown memory file, you need to detect and collapse structural duplicate rows by consolidating multi-block section content into a single canonical version. This prevents sections from drifting into multiple competing versions over repeated memory-flush cycles.

Autonomous agent memory flush operations create duplicate content blocks due to unattended reflect and flush drift. This happens when the process prepends repeated blocks under headings like Recent Articles and Skills Built, causing structural duplicates to accumulate over repeated cycles without manual intervention.

To consolidate duplicate markdown headings without losing data, select a canonical block based on completeness rules, such as the most complete counts. Fold any unique information from non-canonical blocks into the canonical version before rewriting only the affected sections.

Yes, you can automatically clean up structural memory duplicates in unattended agent workflows. The process detects drifted single-canonical sections, rewrites them, runs scripts/memory-dedupe --fix to eliminate re-exposed pointer duplicates, and logs results while optionally notifying when drift is fixed.

The best way to merge multiple competing versions of a markdown section is through intelligent merge and rewrite. This approach selects the most complete canonical block, folds in unique information from non-canonical blocks, and rewrites only the targeted drifted sections to maintain memory cleanup consistency.

Running a memory deduplication script fixes both structural blocks and pointer references. After structural rewrites consolidate the content blocks, the companion pointer dedupe runs scripts/memory-dedupe --fix afterward to eliminate any re-exposed pointer duplicates, keeping the memory cleanup consistent across the file.