New Claude 🤝 Entire. You can now use Claude Fable 5.2 to create custom review agents in Trails, our alternative for the pull request that puts confidence evals at the center of review.
Your agents write most of your code. So why don’t they have their own accounts that track changes clearly under their own name?
Now, on Entire, they do. Configure a scoped service account for your project and assign a role to your agent, the same way you would for a human collaborator. More in @HaimantikaM’s blog post:
Not every file in your repo needs the same agentic review. Path filters for Trails let you choose which changes trigger which reviews, cutting unnecessary runs, review noise, and token spent.
Learn more in @lizziepika ’s first Trail guide and never walk alone.
“You start a pull request and you have it review and it finds 3 things, and then you address those 3 things and now you feel like you're good. You're wrong.”
A sneak peak of our first Entire Espresso with @GergelyOrosz and our CEO @ashtom.
If you missed the event and want to get more of the talk they had over espresso and mochi donuts, watch it here:
What comes after the pull request? A great question for breakfast. Our CEO @ashtom and @GergelyOrosz , author of @Pragmatic_Eng newsletter, had a talk at our first Entire Espresso in San Francisco. October 29, over coffees and mochi donuts. ☕