Urgent.News

What's breaking now, across thousands of outlets.

AI

CORE Is a Product. I Still Won't Call It Production-Ready. Help Me Break It.

Three weeks ago I wrote about a 72-hour autonomous run: 68,495 blackboard entries, zero restarts, about thirty-five workers. The system stayed alive from beginning to end. And I refused to sign it off. The reason was simple: the evidence proved continuity. It did not prove the thing the gate actually claimed — that CORE's autonomous loop was reliable. So G4 stayed red. That story has now…

Three weeks ago, the author wrote about a 72-hour autonomous run of CORE with no restarts and 68,495 blackboard entries. However, the system's autonomous loop was not deemed reliable enough to sign off on, keeping G4 red. After further investigation, the author found that the high finding count was mostly due to churn and noise. Upon separating the noise from actual governance violations, only nine real deterministic remediation proposals ran during the week, all completing successfully.

While the engine itself works, the question now is whether CORE is truly a product or still not ready for production. The author lists several product features CORE now possesses, such as released versions, public installation paths, PyPI packaging, containers, a GitHub Action, CLI, upgrade and database migration paths, public documentation, deprecation behavior, onboarding for repositories, governed execution, rollback, and explicit failure states.

However, some implementation details still need to be fixed, such as documented API routes not being available at the documented versioned path and a pack causing audit crashes due to duplicate rule IDs.

The author emphasizes that CORE's autonomous loop reliability gate remains unmet and that the system must have explicit failure states for findings to end somewhere other than silently dropped or endlessly reposted. Currently, CORE's production readiness status is "NOT ATTESTED" with only two gates met and twelve partial. The author remains uncertain if someone other than them can operate the system properly.

Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at dev.to →

More in AI

OpenJDK Banned AI-Generated Code. Here Is Exactly What the Policy Allows

The story hit the Hacker News front page again this week: Oracle bans AI-generated code from OpenJDK. 536 points, 381 comments, and a comment section full of people arguing about a policy most of them…

  • OpenJDK bans AI-generated code entirely.
  • Policy cites reviewer burden, safety, and IP concerns.
  • GraalVM permits AI assistance with final user-authored output.

27 days of autonomous agents: nothing crashed, disk just hit 85%

I did not plan to write about a hard drive today. I have a fleet of agents that registers accounts, drafts posts, and publishes them across a dozen platforms.

  • Autonomous agents ran autonomously for 27 days without crashing
  • Disk usage hit 85% despite no errors in system logs
  • Failure was due to disk space running out, not system crash

When an AI agent says it's done and it isn't

The edit is usually fine. The claim about the edit is the problem, and it is a harder one. You ask for a change across four files.

  • AI agents can claim tasks completed without verification
  • Language models generate plausible continuations, not repository state
  • Testing tool's actions distinguishes between running and summarizing

More from Saturday 3 October →