Urgent.News

What's breaking now, across thousands of outlets.

AI

The agent wrote the migration. Are you really going to Approve it?

Tuesday afternoon, that Approve button Tuesday, 3 p.m. A PR notification lands in Slack. Junior writes in the channel: "Cursor generated this migration for me — can you take a look?" You open the diff. Eighty lines of ALTER TABLE . Add columns, change types, add indexes, attach foreign keys — every line is familiar on its own. Syntax is clean. Naming matches team conventions. Mechanically, it all…

The agent autonomously generated a migration consisting of eighty lines of ALTER TABLE statements, including adding columns, changing data types, adding indexes, and attaching foreign keys. While the syntax appeared clean and the naming conventions matched the team's standards, there were a few concerns that needed attention. Specifically, the NOT NULL constraint on biz_order lacked a default value, the type change on sys_user would lock the table, and the down migration might not accurately restore the previous schema.

The PR description merely stated "Agent self-review: should be fine," which was not sufficient evidence. Approving this migration without proper scrutiny could lead to production issues, as postmortems would seek to determine who approved the flawed version.

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

The failures that don't fail loudly

I spent a week building an agent that upgrades dependencies and repairs what the upgrade breaks. Dependabot opens the PR and walks away; this one stays until the tests are green.

  • Scanner failed to detect vulnerabilities in lockfile
  • Agent reported 124 vulnerable packages instead of 3
  • Multiple scanner under-reporting vulnerabilities

Three layers of automated fact-checking for an LLM newsroom (and the bugs that forced each one)

Our site, presentofai.com , publishes AI industry analysis daily with no human in the writing loop: agents ingest news and company filings into an event timeline, score them, and synthesize digests…

  • First layer, article-level critic checks drafts against source events for errors
  • Second layer, search-verified claim checking extracts and verifies crucial claims
  • Third layer, data auditing re-reads timelines for recurring failure classes

More from Sunday 30 August →