Today we’re expanding Detection Mode for code review so that you can choose the ideal performance profile for each review:
Ultra: Frontier bug detection for the most consequential reviews. Leverages a much higher number of parallel agents using more advanced coding models. Useful when maximizing bug detection matters more than cost.
Balanced (default): Industry leading high bug detection. Recommended for teams who use Agents to review PR comments, where the inconvenience of occasional false positives is minimal.
Precise: Similar bug detection as Balanced mode, but higher confidence findings due to additional agentic validation to reduce noise from false positives and low-value findings.
Budget: Fast, low cost bug detection.
We’re all accustomed to choosing different intelligence-level models for different tasks… the same should be true for code review.
I’m particularly proud of this launch because Macroscope’s Balanced pipeline is already exceptional and (according to the latest independent benchmarks from Martian), has the highest f1 score of any other code reviewer. Our new Ultra pipeline is a further 10% improvement on this!
We just launched a new version of @Macroscope code review that finds significantly more bugs, including 71% more critical issues, 27% more high-severity issues, 50% more medium-severity issues, and 31% more valid issues overall in our benchmarks, at the expense of a modest increase in false positives and review latency.
Ask Macroscope Anything (AMA) is now Macroscope Agent, and it got a big upgrade today:
→ 6 new integrations (Sentry, Amplitude, GCP Logging, GitHub API, image gen, and MCP)
→ Event-driven automations triggered by commits and PRs
→ Webhook delivery + an API
Deep codebase understanding means an agent that (without being told):
> knows your feature branch exists
> knows the bug lives there
> targets the fix correctly
Fix bugs from anywhere with Macroscope
Today we're releasing Macroscope Code Review v3. Based on our internal benchmarks:
→ Detects up to 3.5x more bugs that would cause real production damage (data loss, security breaches, crashes, etc). The kind you'd block a PR over.
→ Precision increased to 98% (up from 75%) which means significantly fewer false positives.
→ Leaves 22% fewer comments overall, including 64% fewer nitpicks in Python and 80% fewer in TypeScript.
Hello World 👋
Macroscope uses your codebase to automatically summarize product development activity, answer questions (about product and codebase functionality, and dev progress) and catch bugs.
Sign up now and get a 2 week free trial.
@CicmilJovan I taught a bootcamp for 8 years. Only my last cohort used AI assistants. Many of their final projects were completely unusable because of all the vibe coding going on. LLMs only help devs who know what they’re doing. And even then, grain of salt heavily applies.
@joe_gichuru@housecor That’s a bit of an assumption. True, I don’t use TS every day in my 9-5, but I’m not a novice. I actually taught it to 100’s of students.
My original point was to agree with Cory’s assertion that excessive typing can slow us down.
@mfpears@housecor Yes. But only for a few weeks. And like I said it’s an existing codebase. So it’s not like I’ve had much time to absorb a lot of it. It’s a great app, but lots of abstractions to the point where it makes it difficult to navigate the code base.