Hacker News Reader

The WAL mode fix for concurrent MCP clients is clever.

One thing I'm curious about — how are you handling memory relevance decay? If I stored context about a project six months ago that's no longer accurate, the hybrid search might still surface it confidently. Do you have any mechanism for staleness detection or does the user have to manually curate old memories?

View on HN

Staleness isn't solved yet - honest answer is manual curation for now. age_days is on every memory so clients can factor it in, but there's no automatic decay.

It's on the roadmap. I wanted retrieval solid before layering temporal complexity on top. Longer term — age-weighted scoring, surfacing "this is 6 months old" in the MCP response metadata.

For the moment there is scoping feature that allows you to store & filter which helps narrow down retrieval.

Reply on news.ycombinator.com