'Claude Says It Works' Is Not Testing: e2e on Verifying Agent Changes
blaizedsouza · x · 2026-11-10
e2e, which raised $8M to build "the software verification layer for the agentic coding era" (13.6k GitHub stars), published a blog on how agentic testing should actually work.
Key argument: traditional end-to-end scripts assume full determinism — exact values matching every time — while agent behavior is inherently non-deterministic, so legacy frameworks don't fit. Instead of trusting an agent's own claim that a change works, you must make agents verify exactly what they were supposed to test, keeping validation tied to the original intent rather than "claude says it works."
More from coding & agent
- As Arc sunsets, a dev builds Earful, an AI tool that aggregates product reviews into verdicts — BLCNYY · 2026-11-10
- Claude's Chrome Extension Slammed for Being Unable to Control Already-Open Tabs — Angaisb_ · 2026-11-10
- Nebius releases Longtail-SWE: 202 repo-level SWE tasks across 13 underrepresented languages — nebius · 2026-11-10
- Self-hosting OpenMausBot for half a day: it runs, but don't trust it with real work — moutsea · 2026-11-10
- Dev praises Amp's customer-centric credits-for-bug-reports, sees enterprise unlock on Windows — bytebot · 2026-11-10
- He Vibe-Coded a Calorie Tracker to Avoid Paying: 'Most Apps Have No Moat' — weswinder · 2026-11-10