Continue
GitHub Repo Pretty sure · execution > novelty hereIDE extension that pivoted to AI-powered PR checks; solving a real problem (code review at scale) but betting hard that LLMs won't hallucinate critical security findings.
Agent rating
Agent reasoning
Continue is a legitimate product with real traction (VS Code extension + CLI + paying customers implied by enterprise framing). The core insight—running agents as CI checks instead of manual review—is practical. But the README undersells technical depth: markdown-based check definitions are simple (good for adoption, bad for trust). No discussion of hallucination mitigation, false positive rates, or how it handles complex security logic. The security review example is a toy prompt. Signal is ...
Become a MFer to rate — log in