🔱 PR-Agent — CLI/OAuth subscription fork

A community fork of qodo-ai/pr-agent (Apache-2.0). Not affiliated with or endorsed by Qodo — all credit for the original project goes to its authors and contributors. The full upstream README follows below, unchanged.

What this fork adds

• CLI/OAuth AI handler (ai_handler=cli) — run PR review on your chat/CLI subscription (claude -p for Claude Max, codex exec for ChatGPT Pro) instead of an API key. No API key, no GitHub Copilot quota — it runs on the subscription you already pay for.
• Local MCP server (pr-agent-mcp) — review a PR from inside any MCP window (Claude Code / Cursor / Codex), with cross-model review: window = Claude, reviewer = ChatGPT (or vice-versa) → an independent second opinion.
• Reusable GitHub Action — run the reviewer in CI on a subscription token.
• Inline /improve by default — committable, line-anchored suggestions.
• Security hardening of the CLI handler — executable allowlist (RCE defense), timeout subprocess kill, robust config parsing.
• Slim base install — heavy cloud SDKs (AWS/Azure/GCP) are now optional extras (pr-agent[aws] / [azure] / [gcp] / [clouds]); the base stays light for the common GitHub + OpenAI/Anthropic (or subscription) path.

Fork documentation

Doc	What it covers
docs/oauth-cli-mode.md	run on a CLI/OAuth subscription (no API key) — enable, CI, output behavior, security
docs/mcp-local.md	local MCP server — Quickstart to use it in a Claude Code window, cross-model
docs/how-it-works-explained.md	code-grounded legend of the flow + where the CLI/OAuth handler plugs in
docs/orchestrator-integration.md	PR-Agent as a review producer behind an orchestrator (vs. an in-house reviewer)
CHANGES.md	the full, itemized list of fork changes (Apache-2.0 §4(b))

Quickstart (Claude Code): install with pipx install "pr-agent[mcp] @ git+https://github.com/jailtoncarlos/pr-agent.git", then follow the Quickstart in docs/mcp-local.md.

---

Everything below this line is the upstream qodo-ai/pr-agent README, preserved as-is — except the How It Works section, whose external image was replaced with a versioned mermaid diagram (see docs/how-it-works-explained.md).

---

The Original Open-Source PR Reviewer

---

This repository contains the open-source PR Agent Project. It is not the Qodo free tier.

Try the free version on our website.

👉Get Started Now

PR-Agent is an open-source, AI-powered code review agent and a community-maintained legacy project of Qodo. It is distinct from Qodo’s primary AI code review offering, which provides a feature-rich, context-aware experience. Qodo now offers a free tier that integrates seamlessly with GitHub, GitLab, Bitbucket, and Azure DevOps for high-quality automated reviews.

Big News for PR-Agent

PR-Agent has a new home!

After years of building this tool alongside the community, Qodo has donated PR-Agent to the open-source community - and we couldn't be more excited about what comes next.

The project now lives in the PR-Agent org on GitHub, is fully community-owned, and is open for contributions and additional maintainers.

What else changed:

• Docs moved to - www.pr-agent.ai
• Qodo Merge (Qodo 1.0), the hosted URL, which was the enterprise version of PR-Agent, has been rebranded and evolved into Qodo (Qodo 2.0), a full AI code review platform.

Table of Contents

• Getting Started
• Why Use PR-Agent?
• Features
• See It in Action
• How It Works
• Data Privacy
• Contributing

Getting Started

Note

Docker Hub namespace migration. Releases 0.34.2 and later are published under pragent/pr-agent. Older releases (up to and including v0.31) remain available at the legacy codiumai/pr-agent namespace as a frozen archive — no new images are pushed there. Update any pinned image: / docker pull / uses: docker:// references when upgrading to 0.34.2+.

🚀 Quick Start for PR-Agent

1. GitHub Action (Recommended)

Add automated PR reviews to your repository with a simple workflow file:

# .github/workflows/pr-agent.yml
name: PR Agent
on:
  pull_request:
    types: [opened, synchronize]
jobs:
  pr_agent_job:
    runs-on: ubuntu-latest
    steps:
    - name: PR Agent action step
      uses: the-pr-agent/pr-agent@main
      env:
        OPENAI_KEY: ${{ secrets.OPENAI_KEY }}
        GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

Full GitHub Action setup guide

2. CLI Usage (Local Development)

Run PR-Agent locally on your repository:

pip install pr-agent
export OPENAI_KEY=your_key_here
pr-agent --pr_url https://github.com/owner/repo/pull/123 review

Complete CLI setup guide

3. Other Platforms

• GitLab webhook setup
• BitBucket app installation
• Azure DevOps setup

Why Use PR-Agent?

🎯 Built for Real Development Teams

Fast & Affordable: Each tool (/review, /improve, /ask) uses a single LLM call (~30 seconds, low cost)

Handles Any PR Size: Our PR Compression strategy effectively processes both small and large PRs

Highly Customizable: JSON-based prompting allows easy customization of review categories and behavior via configuration files

Platform Agnostic:

• Git Providers: GitHub, GitLab, BitBucket, Azure DevOps, Gitea
• Deployment: CLI, GitHub Actions, Docker, self-hosted, webhooks
• AI Models: OpenAI GPT, Claude, Deepseek, and more

Open Source Benefits:

• Full control over your data and infrastructure
• Customize prompts and behavior for your team's needs
• No vendor lock-in
• Community-driven development

Features

PR-Agent offers comprehensive pull request functionalities integrated with various git providers:

		GitHub	GitLab	Bitbucket	Azure DevOps	Gitea
TOOLS	Describe	✅	✅	✅	✅	✅
	Review	✅	✅	✅	✅	✅
	Improve	✅	✅	✅	✅	✅
	Ask	✅	✅	✅	✅
	⮑ Ask on code lines	✅	✅
	Help Docs	✅	✅	✅
	Update CHANGELOG	✅	✅	✅	✅
USAGE	CLI	✅	✅	✅	✅	✅
	App / webhook	✅	✅	✅	✅	✅
	Tagging bot	✅
	Actions	✅	✅	✅	✅
CORE	Adaptive and token-aware file patch fitting	✅	✅	✅	✅
	Dynamic context	✅	✅	✅	✅
	Fetching ticket context	✅	✅	✅
	Interactivity	✅	✅
	Local and global metadata	✅	✅	✅	✅
	Multiple models support	✅	✅	✅	✅
	PR compression	✅	✅	✅	✅
	Self reflection	✅	✅	✅	✅

---

See It in Action

/describe

---

/review

---

/improve

---

How It Works

The diagram below is PR-Agent's flow — a versioned, faithful recreation of the upstream "How It Works" figure. The AI steps (PR-Agent Planning and every tool) run through a pluggable AI handler: that is the single seam where this fork adds ai_handler=cli to run on a chat/CLI subscription (OAuth) instead of an API key. For a step-by-step legend grounded in the code — and where the handler plugs in — see How It Works — explained.

flowchart LR
    A1["In Public GitHub PR\ncomment: @CodiumAI-Agent [/command]"]
    A2["In Private GitHub/GitLab/BitBucket PR\nafter installation, comment: [/command]"]
    A3["Run PR-Agent locally with CLI"]

    subgraph DIGEST["PR-Agent Digest Request"]
        D1["PR status detection"]
        D2["Hunks detection & prioritization"]
        D3["detect contribution.md / guidelines.md\n(FUTURE)"]
    end

    subgraph PLAN["PR-Agent Planning"]
        P1["Token-aware compression & prioritization"]
        P2["Analyze user request"]
        P3["Questioning user"]
        P1 --> P2
        P2 -->|"/reflect"| P3
    end

    ROUTER{"Which function?"}

    T1["Similar issue finder tool"]
    T2["Reviewing tool"]
    T3["Auto-description tool"]
    T4["Q&A tool"]
    T5["Labeling tool"]
    T6["Code-suggestion tool"]
    T7["Changelog tool"]
    T8["Code-docstring tool"]
    T9["Future tool"]

    R1["PR comment"]
    R2["PR comment"]
    R3["PR description"]
    R4["PR comment"]
    R5["PR labels"]
    R6["PR inline code suggestions"]
    R7["Update changelog"]
    R8["PR inline code suggestions"]

    PC["PR comment"]
    UR["User response"]

    A1 --> DIGEST
    A2 --> DIGEST
    A3 --> DIGEST

    D1 --> P1
    D2 --> P1
    D3 -.-> P1

    P2 --> ROUTER
    P3 --> ROUTER

    ROUTER -->|"/similar_issue"| T1
    ROUTER -->|"/review"| T2
    ROUTER -->|"/describe"| T3
    ROUTER -->|"/ask"| T4
    ROUTER -->|"/generate_labels"| T5
    ROUTER -->|"/improve"| T6
    ROUTER -->|"/update_changelog"| T7
    ROUTER -->|"/add_doc"| T8
    ROUTER -->|"/command"| T9

    T1 --> R1
    T2 --> R2
    T3 --> R3
    T4 --> R4
    T5 --> R5
    T6 --> R6
    T7 --> R7
    T8 --> R8
    T9 -->|"..."| PC

    R1 --> PC
    R2 --> PC
    R3 --> PC
    R4 --> PC
    R5 --> PC
    R6 --> PC
    R7 --> PC
    R8 --> PC

    PC --> UR
    UR -.-> DIGEST

Loading

_{The original upstream diagram image lived at qodo.ai/images/pr_agent/diagram-v0.9.png; this fork replaces it with the versioned, faithful mermaid recreation above. The AI handler is not a node here — it is the LLM backend used inside PR-Agent Planning and every tool; see the explained doc for where ai_handler=cli (this fork) sits.}

Data Privacy

Self-hosted PR-Agent

• If you host PR-Agent with your OpenAI API key, it is between you and OpenAI. You can read their API data privacy policy here: https://openai.com/enterprise-privacy

Contributing

To contribute to the project, get started by reading our Contributing Guide.

❤️ Community

This open-source release remains here as a community contribution from Qodo — the origin of modern AI-powered code collaboration. We’re proud to share it and inspire developers worldwide.

The project now has its first external maintainer, Naor (@naorpeled), and is currently in the process of being donated to an open-source foundation.