session-compression

Session compression extracts key decisions, context, and open items into memory artifacts, updating per-topic memory files with YAML frontmatter and a MEMORY.md index. Running it at the end of a multi-file refactor ensures future sessions start with full context.

Session compression detects changed files via git diff and git status, then distills the architectural choices and reasoning behind those code changes. It consolidates the extracted decisions into structured memory files for long-term knowledge management.

Session compression consolidates or merges memory entries by updating existing per-topic memory files rather than creating duplicates. It adjusts the MEMORY.md index to reflect additions, removals, and unresolved open items during the compression process.

Session compression applies directly to end-of-session workflows across code refactors, architecture decisions, and multi-file project work. It preserves what changed, why it mattered, and which tasks remain open for future reference.

A compression report summarizes the created and updated memories generated during session compression. It provides an overview of the structured memory artifacts extracted, ensuring you can verify which decisions and open items were persisted.