IntentLint

Semantic CI

Lint your code
with prose

Write engineering invariants in English. IntentLint evaluates them on every pull request and turns high-confidence violations into ordinary CI checks.

Free 2,000 evaluations to start!

Install on GitHub See examples

IntentLint evaluating an English rule against a pull request and marking it clean.

Cost

It costs pennies to lint your pull requests against all of your written rules.

Speed

Our prose-based linters all run in only a few seconds. Give your engineers, and your coding agents, immediate feedback.

Share tribal knowledge

Check into version control all of the critical tribal knowledge that is important to your team and your codebase.

Beyond normal linters

Standard linters are executed as code, limiting their ability. Prose is not constrained by AST parsers.

Not a general AI reviewer

IntentLint asks one question: does this PR violate a named rule? Silence is success. Shadow mode first, then promote rules in .intentlint.yml.

One GitHub Check

High-confidence violations land as one check on the pull request. Engineers see a named rule, not a thread of suggestions.

Configuration

- id: jobs/idempotent-workers
  description: >
    Background workers that make external
    side effects must be idempotent.

Pull request

def perform(order_id):
    order = Order.find(order_id)
    payment_provider.capture(order.card)
    order.update!(charged: true)
✕ jobs/idempotent-workers

This worker is not idempotent. The external charge happens before the local record is marked as charged. A retry after a successful capture can charge the customer twice.

99.2% confidence