autter.config.yml at the root of your repository using plain English descriptions — no regex expertise or complex YAML syntax required.
Rule types
Autter organises rules into three categories that map to the most common sources of problems in pull requests.Security rules
Security rules detect vulnerabilities and enforce safe access patterns across your codebase. Examples from Autter’s built-in ruleset:Detect security vulnerabilities— scans for common vulnerability patterns across all changed filesNo Direct process.env Access— environment variables must be accessed through the config module, not read directly
Architecture rules
Architecture rules enforce the structural patterns your team has agreed on — things like which modules are allowed to import from which, or how configuration must be accessed. Example: Preventing directprocess.env access enforces the pattern that all environment variable access flows through a single validated config module, making configuration errors easier to catch at startup rather than at runtime.
Reliability rules
Reliability rules encode the patterns your team has established to prevent production failures. Example:Require error boundaries — ensures React component trees have error boundary wrappers so individual component failures don’t crash the full page.
Configuration reference
Add arules block to autter.config.yml to configure enforcement. Each key under rules is a rule category name.
autter.config.yml
Rule fields
How strictly Autter enforces this rule category. Accepts
"block", "warn", or "info".block— prevents merge until the issue is resolvedwarn— adds a review comment but allows merge to proceedinfo— informational only, no review action required
A list of glob path patterns to exclude from this rule. Files matching any pattern in this list will not be flagged, even if they would otherwise trigger the rule.Use exceptions for known legacy directories or third-party code you don’t own and can’t change.
Convention detection
Autter automatically learns your codebase’s conventions from your merge history and flags deviations — without you needing to document every rule manually. You can inspect the conventions Autter has detected and export them for onboarding documentation:conventions list produces a table of every pattern Autter has learned, its category, and how consistently your codebase follows it:
Convention detection runs automatically. You do not need to configure anything for Autter to learn from your merge history. Run
npx autter init --learn after first connecting a repository to seed the initial convention catalogue from existing merged PRs.