LlamaPReview Support

LlamaPReview is an actively maintained, open-source GitHub pull-request reviewer. The hosted service reviews public repositories only and is free for public repositories. This page explains where to take usage questions, reproducible problems, review-quality feedback, and security reports.

Frequently Asked Questions

How do I install LlamaPReview?

Install the LlamaPReview GitHub App and grant it access to the public repositories you want reviewed.

Does the hosted service review private repositories?

No. After verifying the webhook signature and parsing the minimum visibility fields, the hosted service generically acknowledges and discards a private-repository event. It creates no durable product record, calls no model provider, and makes no GitHub product API call for that event.

When does a review run?

A review runs for an eligible public pull-request event with a valid webhook signature. Opening a non-draft pull request or marking a draft ready for review are the supported entry points. Every admitted run is pinned to the pull request's exact head commit.

How much hosted review capacity is available?

The official hosted service admits up to 3 review runs per repository per UTC day, subject to shared daily capacity and available model budget. These are admission limits, not guaranteed availability. Capacity counts attempts admitted after deterministic skip checks and before the first paid model call, not successfully delivered reviews. Over-capacity requests are skipped, not queued for the next day. The first ordinary pull request over a repository's daily limit may receive one skip notice; later pull requests that day and head successors stop silently. The global circuit breaker is always silent. Read the capacity policy and skip-message contract.

Where is the source, and where should I report a bug?

The active source is JetXu-LLM/LlamaPReview. Use Issues for reproducible bugs and concrete work items. Use Discussions for questions, ideas, and public review examples.

What if a review looks wrong, incomplete, or noisy?

Open a review-quality Issue or Discussion with a public pull-request link, the output in question, and what you expected instead. Share only public or already-redacted evidence; never post private source, provider payloads, installation IDs, secrets, or private logs.

How do I report a security problem?

Follow the repository's Security Policy. Do not file a public Issue for a vulnerability.

Can I self-host LlamaPReview?

Yes. The repository contains a generic AWS reference for the same public-repository-only Webhook and Pipeline path. The reference disables the official hosted quota by default. Self-hosters supply their own GitHub App, model-provider credentials, and AWS account, and pay their own provider and AWS costs. The reference leaves one-time head succession enabled. New source installations default to an OpenRouter GPT-6 Luna max trial with documented qualification limits; the existing DeepSeek profile remains selectable. Direct OpenAI is also an explicit option requiring a stage profile and verified budget policy. These source choices are separate from the hosted service, which uses the OpenAI API directly. The current reference does not include a private-repository mode. Read the self-hosting boundary.

Getting Help