Skip to main content

Glossary

MR or Merge Request​

In GitLab, the request to merge changes from one branch into another. Most Buzzbin capabilities revolve around merge requests.

Repo or Repository​

A GitLab project connected to Buzzbin. In Buzzbin, a repository is the main unit for settings, spend, and review history.

Project Access Token​

A GitLab project-scoped token that Buzzbin uses to validate the repository, register the webhook, clone the repository, and send output back to GitLab.

Webhook​

The mechanism GitLab uses to send events such as opened MRs, new pushes, or MR notes to Buzzbin.

Webhook Delivery​

One actual event delivery that GitLab sends to Buzzbin's webhook endpoint. The Webhook Deliveries page shows whether that delivery was enqueued, skipped, or rejected.

Webhook Health​

The health state of the registered GitLab hook. In Buzzbin it can be UNKNOWN, HEALTHY, DISABLED, or MISSING.

Review Job​

The queued unit of review work. A relevant event can create a review job that the engine processes later.

Review Run​

One actual execution of a review job. A run includes status, timestamps, model, token usage, cost, and final output.

Finding​

A concrete issue or observation identified by Buzzbin during a review. A finding usually includes severity, category, confidence, message, and recommendation.

Severity​

The urgency of a finding. Current levels are INFO, LOW, MEDIUM, HIGH, and CRITICAL.

Category​

The type of finding, such as SECURITY, BUG, PERFORMANCE, or TESTING.

Confidence​

How sure Buzzbin is that a finding is real. Current levels are HIGH, MEDIUM, and LOW.

Merge Risk​

An overall estimate of the risk of merging the MR. This is different from the severity of any single finding.

Suppressed Finding​

A finding that was recorded but not published on the MR. This can happen because of low confidence or publishing rules.

Backfill​

The feature that reviews a selection of currently open MRs after a repository is connected, so you do not have to wait for new events.

Review Config​

Buzzbin settings that control which capabilities are enabled, which categories are allowed, which model is used, and what language the output should use.

Review Model​

The model used for Code Review and some related tasks such as title and description generation.

Answer Model​

The model used for Comment Answers, often chosen to be faster or cheaper than the review model.

Daily Run Cap​

The per-repository limit on how many review executions can run in one day. This is one of the usage guardrails.

Wallet​

The organization's prepaid credit wallet that review costs are deducted from.

Available Balance​

The part of the wallet balance that is currently spendable.

Held Balance​

The portion of credit reserved for in-flight runs that has not been fully settled yet.