MCP server that timestamps every web extraction for AI agents

LLMs hallucinate recency. They'll tell you a project is active when it hasn't been touched in two years, or cite outdated docs as current. The problem isn't the content — it's that there's no timestamp on when it was retrieved.

FreshContext wraps every extraction in an envelope with retrieved_at, content_date, and freshness_confidence fields. The agent always knows when it's looking at data, not just what it says.

Ships with 7 tools: GitHub repo extraction, Hacker News sentiment, Google Scholar, YC startup scraper, GitHub repo search, npm/PyPI package trends, and extract_landscape — one call that queries all sources simultaneously and returns a unified competitive intelligence report.

Works locally (Playwright) and on Cloudflare Workers (Browser Rendering API). MIT license.

Would love feedback on the FreshContext envelope pattern specifically — is this a useful primitive for agent reliability?