Popular News
Latest News
Top Stories
View all
Reviewing agent code is a trust problem, not a diff problem
From Mexico: JetBrains Research and Lund University say reviewing what a coding agent wrote isn’t about seeing what changed. It’s about knowing where to look hard. Here’s their three-level review idea and what I’d steal from it today.
GitHub is rebuilding Git for repos full of agents
From Mexico: GitHub’s Oct 6 engineering post says agent fleets are breaking the old trade-off in its Git storage, so it’s splitting durable storage from compute and coordinating only the ref update. Here’s what the numbers mean for small teams.
GitLab says coding agents have to earn their autonomy
From Mexico: GitLab published its Security Standard v1.0 on Oct 6, a five-stage ladder that coding agents climb before they can merge on their own, on the same day Transcend pushed /goal flows that stop waiting for a human.
Obelisk 0.42 keeps agent progress in the database, not the process
From Mexico: Obelisk 0.42 (Oct 4) stores coding-agent workflow progress in a database, wraps generated code in nested server/app/deployment grants, and ships a workflow-agent prototype that can deploy and inspect apps after the process dies.
Cloudflare gives the agent a search box instead of a guessed URL
From Mexico: Cloudflare’s Oct 2 post puts web search on AI Gateway, so an agent can ask Ceramic, Exa, or Linkup instead of guessing a URL and curling a 404. Native server tools are still coming soon. The docs call it an open beta.
Supabase puts the backend in the repo the agent already edits
From Mexico: Supabase’s Oct 2 post moves schema and project config into the repo, runs a local stack without Docker, and lets an app ship its own MCP server. Compute, for long jobs, is still a private alpha.
Pi 1.0 keeps the coding agent small and finally takes MCP
From Mexico: Earendil shipped Pi 1.0 on Oct 1, a minimal coding-agent harness that adds MCP through Codemode, deferred tool loading, and virtual models. They kept the terminal agent small and put long runs in a separate experiment.
Sentry’s Seer Agent can change the project, one chat at a time
On September 29, Sentry’s Seer Agent can create dashboards, alerts, and triage inside a project, and it can read Datadog or Google Cloud while it diagnoses. The write permission does not outlive the chat.
Momentic’s Mo bug-bashes the app instead of leaving you a suite
Momentic launched Mo on September 28: give it a URL or a mobile build, and a swarm of agents bug-bashes the app. You get a report with recordings and repro steps. A lasting test suite is optional, and the beta counts are Momentic’s own.
Autoheal’s $7.9M seed targets the work after coding agents commit
Autoheal announced a $7.9M seed on Sep 28 for a self-improving software factory: shared context for post-coding agents, plus an evaluator and a healer that only land changes through a pull request you approve.
