I’ve been building something called FreshContext. The core idea is simple: AI systems often treat fresh and stale retrieved information as equally useful. FreshContext is an attempt to fix that. Instead of only retrieving information, the system applies temporal scoring before signals reach an LLM or agent workflow. That means: source timestamps matter decay matters provenance matters retrieval timing matters What I built so far 1. FreshContext MCP A Cloudflare-native MCP server with 21 tools focused on freshness-aware retrieval and live intelligence workflows. Current stack: Cloudflare Workers D1 KV structured JSON envelopes freshness scoring observability tooling GitHub: https://github.com/PrinceGabriel-lgtm/freshcontext-mcp 2. Fresh HN Feed A freshness-ranked Hacker News signal feed. Instead of simply listing posts chronologically, the feed scores signals using temporal decay and relevance weighting. GitHub: https://github.com/PrinceGabriel-lgtm/fresh-hn-feed 3.…