How Threadline Works

An AI-powered linter based on your natural language documentation. Define your standards as rules or principles - your choice.

The Flow

Define Your Threadlines

Create a /threadlines folder in your repository. Add markdown files ('threadlines') defining your team's coding standards. Each file is a single code quality standard or convention.

Example: threadlines/feature-flags.md

# Feature Flag Standards

All feature flags must:
- Use our FeatureFlagService (not direct SDK calls)
- Include fallback behavior for when flags are unavailable
- Log flag evaluations to our analytics service

Your threadlines live with your code, get version-controlled, and evolve with your code base.

Run the Check

The npx threadlines check command runs in a few seconds from your local CLI and CI pipeline, validating your changes against your team's coding standards.

npx threadlines check

That's it. One command.

Get Focused Feedback

The command returns simple status for each coding standard: compliant, needs attention, or not relevant. You only see the threadlines that require attention - no noise, just actionable feedback.

25 threadlines checked

8 not relevant

16 compliant

1 attention

Feature Flag Threadline

src/api/users.ts:23 - Using direct SDK call instead of FeatureFlagService

src/components/Header.tsx:67 - Missing fallback behavior when flag is unavailable

Fix the issues, run threadline check again, and iterate until everything passes.

Why It Works

Focused AI Checks

Each threadline checks one thing. No AI trying to be everything to everyone.

Parallel & Fast

All threadlines run simultaneously. Get comprehensive feedback in seconds.

Fully Auditable

Every check is logged. Trace exactly what was reviewed and why.

Simple Configuration

Works out of the box. Just add your threadlines and run.