Bun Eyes Image Library, Adversarial Tests

This reads like a security-first roadmap item, not a casual feature add: image handling is one of the fastest ways to inherit parser bugs, crashers, and memory-safety headaches.

• –Mining prior GitHub issues and CVEs is the right move because image libraries have a long history of nasty edge cases that only show up under fuzzing
• –Adversarial tests matter here more than happy-path benchmarks; the real risk is corrupted inputs, odd metadata, and format-specific decoder bugs
• –If Bun adds this surface area, the win is not just convenience but a tighter developer story for apps that already use Bun for CLIs, servers, and AI tooling
• –The tradeoff is maintenance cost: every new codec or transform expands the native attack surface and the compatibility burden
• –The differentiator, if shipped well, will be robustness under hostile inputs rather than just a thin wrapper around existing native image libraries