CHANGES
Read the changes that actually matter.
AI now writes code faster than developers can review it. An agent hands you thousands of lines in a diff, and you have no idea which files could introduce a regression. Panolayer scores every changed file by risk and explains why, so you know which ones to review first. Commit and push from the same place when you’re ready.
Nobody reads a 3,000‑line diff. We all scroll to the end and commit.
Most of a big diff is lockfiles, config, and small edits that don’t need your attention. The few files that could actually break something are buried somewhere in the middle. Panolayer pulls them to the front, so you look there first and keep moving at the speed of the agent.
01
See where to look first
You don’t have time to read every file an agent touched, and you don’t need to. A summary shows you how many files changed, lines added and removed, and every changed file gets a risk score from 0 to 100% with a one-line reason why. The Risk Spectrum puts the highest-risk files at the red end. Risk Bars and Overlap show the same scores vertically, or grouped by what each change touched.
- UI
- Logic
- Data
- API
- Config
- Infra
- Deps
- Security
- Test
- Docs
- Style
- Analytics
- Other
02
Open the file
When a file ranks high, you want to know why before you read it. Click it for the diff, split view or inline, with the full reasoning for its score above it.
03
When the score isn’t enough
The Analyze all feature reads the full change set and returns a summary, an overall risk, specific concerns, and a note on each file. For anything more, Chat opens a conversation with Panolayer Agent, with the changed files and the change itself already attached, so you can ask about it instead of reading it line by line.
04
Compare with your recent commits
A high score means more when you know what normal looks like for your repo. History lists your working tree above your recent commits, each with a risk label, author, and hash. You can tell whether this change is routine or the riskiest thing you’ve committed this week.
05
Commit and push
Once you’ve reviewed what matters, commit from the same screen. Panolayer drafts a commit message from the diff, you can edit. Commit writes to your local branch, and Commit & Push sends it to the remote it tracks. Commits include every changed file. For partial staging, you can use Git directly.
Pull request review happens on your Git host, after this point.
Panolayer holds agents, not you
An agent can’t merge its own work into your branch without your approval. Its changes are verified in their worktree first, and then you decide whether to merge. Your own work is never gated. Commit whenever you want, even with checks failing. The failures stay listed in Verify, so nothing is hidden.
Questions
What does the risk percentage mean?
It’s a model’s read of how much attention a file’s change needs, from 0 to 100%, based on the diff. Scores roll up into Safe, Low, Caution, or Critical.
Is this the same as verification?
No. The score reads the diff. Verification checks the change against the architecture and rules Panolayer derives from your code, and it runs when you save a file and when an agent finishes.
What’s the difference between the score and Analyze all?
The score is a quick read of each file’s diff, on every change. Analyze all reads the whole change together when you ask for it. For deeper questions, Chat hands the change to Panolayer Agent.
Can I commit when checks are failing?
Yes. Panolayer holds agents, not you. The failures stay listed in Verify.
Can I commit only some of my changes?
A commit from Panolayer includes every changed file. To commit a subset, stage it with Git directly.
Does this connect to GitHub?
Commit & Push sends your branch to whatever remote it tracks. Pull requests and review happen on your Git host.