Code Refactoring Context Restore

This handles the messy problem of reconstructing project context when you're jumping back into a codebase or handing off between AI agents. It uses semantic vector search to pull relevant context from your project history, ranks components by relevance, and rehydrates what matters within your token budget. You get full or incremental restoration modes, three-way merge strategies for conflicts, and filters to avoid loading stale decisions. The implementation focuses on multi-agent workflows where preserving decision trails actually matters. It's overkill for solo weekend projects, but if you're managing long-running work across multiple sessions or agents, the semantic ranking and token budget controls are genuinely useful. Requires a vector database setup to work properly.