Documentation

Jargons is an AI code review agent. Once connected, it reviews every pull request in your repositories, posts findings as a review comment, and can open a pull request that applies the suggested fixes. It also scans whole codebases on demand.

1. Sign in with GitHub. 2. Install the Jargons GitHub App and choose which repositories it can access. 3. That's it — Jargons begins watching those repositories for pull requests.

When a pull request is opened, reopened, or updated, GitHub notifies Jargons. Jargons fetches the diff, analyzes it, and posts a review listing each finding with its severity, file location, and a suggested fix. If the findings are fixable, it also opens a companion pull request that applies them — which you can review and merge, or close.

Beyond pull requests, Jargons can scan an entire repository for bugs and structural risks. From a completed scan you can click Open fix PR with Jargons to turn the scan's suggestions into a pull request against your default branch.

In workspace settings you control what Jargons does: pull request reviews, security findings, and codebase scans can each be toggled on or off. Changes apply to every repository in the workspace.

The Agent Health page shows how the review agent is performing: run counts, success rate, latency, throughput, and token/cost usage. Jargons instruments the agent with OpenTelemetry and ships traces and metrics to SigNoz. Core metrics (counts, latency, throughput) are read from your database, while token and cost figures come live from SigNoz.

Need help? Reach out through the Jargons GitHub App page.