The words behind an AI-built change, in plain language.
Sixteen terms in two or three sentences each, enough to make sense of any pull request, plan or review in ProPR. Each entry links to where you meet it in the product, and Git and GitHub terms link to GitHub's own documentation.
Overview
Git and GitHub
- Repository
-
A repository is the folder that holds a project's code and its full history. GitHub keeps it online so people and tools work on one shared copy. Everything else on this page happens inside one.
In ProPR: You connect a repository once, and every plan, task and pull request after that lives inside it. How the repository sets the rules for every run.
- Issue
-
An issue is a written request in a repository: a bug to fix, a feature to add, a question to settle. It has a number, a title and a conversation. Issues give every piece of work a home you can point to later.
In ProPR: Add the
AIlabel to an issue and an agent picks it up and comes back with a pull request. From an issue to a pull request. - Branch
-
A branch is a separate line of work inside the repository. Changes on a branch leave the main version untouched until someone merges them in, which makes a change safe to try.
In ProPR: Every task gets its own branch, named after the issue number and the model that worked on it. The four stages of one change.
- Commit
-
A commit is one saved step of change, with a short message saying what it did. The history of a project is the list of its commits. GitHub shows what each commit changed, and a commit can be reverted.
In ProPR: Every task record lists the commits it produced, and every follow-up commit carries an undo link. The task record.
- Pull request
-
A pull request proposes merging a branch into the main version. It shows the summary, the changed files, the conversation and the checks in one place, and nothing lands until someone merges it.
In ProPR: Every change an agent makes arrives as a pull request, linked to its issue, with the model, time and cost in the description. How to review one when you did not write the code.
- Review
-
A review is a reader's verdict on a pull request: approve it, ask for changes, or leave comments. It takes a few minutes and it is the moment a change gets a second pair of eyes. Reviews are how a team keeps quality high without slowing down.
In ProPR: Comment
/reviewand a second model posts numbered merge blockers, suggestions and a score out of ten, without touching the code. AI review. - Merge
-
Merging brings a branch's commits into the main version. It is the moment a change becomes part of the product. A merged pull request stays in the history with a Revert button, so it can be traced and taken back out.
In ProPR: You press Merge in GitHub, or switch on auto-merge for a plan or task you trust. Keeping a human decision on every merge.
- CI (checks)
-
CI stands for continuous integration: automated checks that run on every commit, such as tests and builds. On a pull request they show up as a list with a green tick or a red cross. Green means the automated checks agree the change is safe to merge.
In ProPR: Switch on automatic CI follow-up and a red check becomes a fix commit on the same pull request, with the failure quoted. Fixing failing checks automatically.
ProPR
- Agent
-
An agent is an AI program that writes and changes code on its own, reading the project and running its tests as it goes. Claude Code, Codex, Antigravity CLI, Mistral Vibe and OpenCode are agents. You pick one, and the subscription or API key you already have pays for it.
In ProPR: ProPR runs the agent you choose in its own isolated workspace and turns its work into a pull request. The five supported agents.
- Plan
-
A plan is a written breakdown of a larger piece of work before any code changes: what to build, in what order, and how to tell it is done. Reading the plan is faster than reading the code. Approve it and the work starts; change it and nothing is wasted.
In ProPR: Describe what you want, refine the draft plan, approve it, and ProPR files one issue per step and starts the agents. Planning.
- Task
-
A task is one run of an agent on one piece of work: an issue, a plan step or a plain instruction. It has a start, a log, a cost and a result, and a small task makes a pull request that is quick to review.
In ProPR: Each task runs in its own workspace, keeps its full log and cost, and ends in a pull request. Implement.
- Goal
-
A goal is an outcome you hand to an agent in a few sentences: the result to reach. The agent works toward it and pauses at checkpoints. You steer it with a comment and check in when you like.
In ProPR: A goal runs on a draft pull request, with checkpoints you follow and an input box to steer it. Goals.
- Epic
-
An epic is a feature too big for one pull request, delivered as several smaller ones. Each piece is reviewed on its own, and the whole lands together at the end, so progress shows in pull requests small enough to read.
In ProPR: An approved plan files one issue per step, runs them in order, and lands each pull request on one epic branch that merges as a whole. Plan an epic and let it land.
- Self-hosted
-
Self-hosted software runs on a computer you control. Your code, logs and settings stay on your machine, and you pay no licence fee. You pick the hardware and the update schedule.
In ProPR: ProPR is free and open source under Apache 2.0, and runs on your own Linux machine with Docker. Self-hosting.
- Hosted
-
Hosted software runs on the vendor's servers, so you sign in and start without installing anything. The vendor keeps it updated and backed up. It is the quickest way to begin.
In ProPR: ProPR Hosted gives you a dedicated instance run by the ProPR team, and your AI usage still goes to your own provider. ProPR Hosted.
- MCP
-
MCP, the Model Context Protocol, is an open standard that lets an AI chat app such as Claude or ChatGPT use outside tools. Connect a tool once and you can ask for things in plain language from the chat.
In ProPR: Connect ProPR to your chat assistant and create tasks and plans, follow them, and read the results without leaving the conversation. MCP from chat.
Put the words to work.
Start with how to review a pull request when you didn't write the code: the two-minute habit that keeps every AI-written change under your control. Then try the live demo, or start on ProPR Hosted.