Appearance
Limits and defaults
Per-repository defaults
The values a repository starts with. All are editable by an admin on the repository's settings page, and most can also be set in a configuration file; see Repository settings.
| Setting | Default |
|---|---|
| Enable AI review | on |
| Trigger on PR events | opened, synchronize (reopened off) |
| Skip draft pull requests | on |
| Exclude patterns | the starting list |
| Post review on GitHub | on |
| Comment severities | warning, critical (info off) |
| Post duplicate findings to GitHub | off |
| Post summary comment | on |
| Max comments per review | on, 20 |
| Spec review | off; available on Pro |
| Default review profiles | none |
Marking a draft pull request ready for review always triggers a review, regardless of the trigger events chosen.
Because info is off by default, minor maintainability findings are kept but not posted inline. Findings held back; by severity, by the comment limit, or because another run already raised them; are counted in the review body and listed in the review detail.
Per-user defaults
| Setting | Default |
|---|---|
| Display mode | choose Technical or Decision view during first sign-in |
Your display mode is yours alone; changing it affects no one else and nothing on your pull requests.
Organization defaults
| Setting | Default |
|---|---|
| Enable review for new repositories | on |
| Data retention (days) | Depends on your subscription plan; review findings and project-status metadata are deleted after the configured period |
Merge confidence bands
| Reading | Score | Meaning |
|---|---|---|
| High | 8.0 and above | safe to merge |
| Medium | above 5.0, below 8.0 | review before merging |
| Low | 5.0 and below | do not merge |
Only completed reviews are rated. One that produced no rating counts as Unrated and is left out of the percentage. See Merge confidence.
The starting exclude list
A repository's exclude list begins with these patterns. Every one of them is editable; remove any entry and the files it matched are reviewed again, and an empty list reviews every file.
text
*.lock
*.min.js
*.min.css
*.generated.*
**/*.d.ts
**/dist/**
**/node_modules/**
package-lock.json
npm-shrinkwrap.json
pnpm-lock.yaml
bun.lockb
go.sum
packages.lock.json
gradle.lockfile
.pnp.cjs
.pnp.loader.mjs
**/.yarn/**
**/build/**
**/out/**
**/.next/**
**/.nuxt/**
**/coverage/**
**/.turbo/**
**/__pycache__/**
**/.venv/**
*.map
*.bundle.js
*.pb.go
*_pb2.py
*_pb2_grpc.py
*.g.dart
*.freezed.dart
zz_generated.*
*_generated.go
**/__snapshots__/**
CHANGELOG.md
*.svg
*.csv
*.sqlite
*.sqlite3
*.sqlite-journal
*.sqlite-wal
*.sqlite-shmThe list covers common lock and dependency metadata, build and coverage output, source maps and bundles, generated code, test snapshots, and selected non-code assets. Lockfiles ending in .lock; yarn.lock, Cargo.lock, Gemfile.lock, composer.lock; are already matched by *.lock. See Excluding files.
Re-runs
| Action | Who | Limit |
|---|---|---|
| Retry review on a failed review | any member | the latest run of an open pull request, when your plan and quota allow a review |
| Re-run the spec review of a completed review | any member | twice per review; does not use review quota |
| Re-run N failed this month on the dashboard | admins | every eligible pull request in one organization whose latest review failed this calendar month (UTC) |
Automations
Automations are available on Pro and always run on the repository's GitHub default branch.
| Limit | Value |
|---|---|
| Repository scope | one repository per automation |
| Frequency | daily or weekly |
| Run window | one hour, in UTC |
| Start-time precision | whole hours only |
| Run deadline | 24 hours, then the run is Timed out |
| Missed window | skipped, not run later |
Moving off Pro pauses every active schedule and cancels scheduled runs that have not started. Returning to Pro keeps the schedules paused until an organization admin enables them again.
Configuration file limits
Limits for .easymerge.yaml. A value past its limit is reported in the configuration-file status bar on Repository settings and ignored; a file past the size or nesting limit is ignored as a whole.
| Limit | Value |
|---|---|
| File size | 64 KB |
| Nesting depth | 10 levels |
| YAML aliases | 50 |
reviews.exclude patterns | 200 |
| Length of a glob | 256 characters |
Wildcard runs per glob (** counts as one) | 3 |
{…} groups per glob | 3 |
comments.max_per_review | 1 to 100 |
reviews.instructions | 4,000 characters |
reviews.path_instructions | 25 entries, 1,000 characters each |
| All instructions together | 12,000 characters |
reviews.path_rules | 50 rules |
Each reviews.filters list | 50 entries |
| A title keyword | 64 characters |
| An author pattern | 64 characters |
| Problems listed per file | 50 |
The advanced keys (language, reviews.verbosity, instructions, path rules and filters) apply only on a plan with advanced review configuration (Pro).