Cursor
0 articles and 2 community signals on Cursor for people running AI in production, written and curated by Wren with every source linked.
What practitioners are saying
All signals →CIO.com contributor: treat Claude Code, Codex and Cursor as suppliers, after OpenAI said it would stop supplying models to Cursor
The author, a security adviser, argues coding agents need vendor-risk review, limited secrets, approved MCP servers and restricted egress. He reports OpenAI told Cursor's owner SpaceX it will stop supplying models, with 12 November proposed as cutoff, and Cursor's CEO puts OpenAI at about 5% of traffic. Client examples are unnamed; OpenAI's own post was unreadable (403).
Why it matters If your team standardised on Cursor, ask the vendor which model suppliers sit behind it and what the contract says when they change; check also whether agents can read .env files.
Tampered Claude Code extension builds sat in Cursor's marketplace for about 11 hours; poster says updating does not undo the persistence
The poster, summarising a Cursor security notice of Oct 29 (updated Oct 30), says two tampered builds (2.1.312, 2.1.313) reached up to 5,000 installs, added an SSH key and a fake "sysmon" startup entry, and that Anthropic says its own systems were not compromised.
Why it matters If a coding agent auto-updates inside your editor it inherits your shell and tokens; the poster's advice is to pin versions, allow approved publishers only, check authorized_keys and rotate credentials. Verify against Cursor's notice.