Open-source AI code reviewer for any GitHub repo. Bring your own LLM key. Drop into a workflow in two minutes. CodeRabbit-class output, zero lock-in.
Erix Buddy is a CodeRabbit-style PR / codebase reviewer that runs as a GitHub Action or locally as a CLI. It produces walkthroughs, mermaid sequence diagrams, severity-tiered findings, suggested patches, SARIF, GitHub annotations, and risk-scored markdown reports β using any LLM provider you have a key for.
This repository hosts the GitHub Action manifest (action.yml), workflow recipes, and end-user documentation. The runtime lives in a separate npm package and is invoked via npx. β @ecodrix/erix-buddy
# 1. Scaffold a review workflow + config in your repo
npx @ecodrix/erix-buddy init
# 2. Set your provider API key as a GitHub secret
gh secret set GEMINI_API_KEY
# 3. Commit and push
git add .github/workflows/erix-review.yml .codereview.json .erixignore
git commit -m "chore: add Erix code review"
git pushYour next pull request gets an automatic AI review.
Want to try it locally first?
GEMINI_API_KEY=β¦ npx @ecodrix/erix-buddy --staged
| π 12 LLM providers | Gemini, Vertex AI, OpenAI, Anthropic Claude, Ollama (local), OpenRouter, Groq, Azure OpenAI, DeepSeek, Mistral, xAI Grok, ECODrix-hosted |
| π Bring your own key | Credentials never leave your machine or your CI runner |
| π PR walkthrough | Title, summary, change table, mermaid sequence diagrams |
| π― Per-file deep review | Severity tiers (critical / major / minor / nitpick / info), 10 categories, confidence, CWE IDs |
| π§ Cross-cutting pass | Architecture / consistency issues that span files |
| π¬ Sticky PR comment | Updates in place on every push, no comment spam |
| π Risk score | Per-file and PR-level (0β100) |
| π Full audit mode | Walk the whole codebase with content-hash cache, file-limit guard |
| π§ Auto-fix patches | Optional git apply of high-confidence fixes |
| π Reports | Markdown, JSON, SARIF (drops into GitHub code-scanning), GH annotations, console |
| π¦ CI gating | Fail the build on critical/major issues |
| π MIT licensed | No telemetry, no phone-home, hackable |
# .github/workflows/erix-review.yml
name: AI Code Review
on:
pull_request:
types: [opened, synchronize, reopened, ready_for_review]
permissions:
contents: read
pull-requests: write
jobs:
review:
if: github.event.pull_request.draft == false
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v4
with: { fetch-depth: 0 } # full history for branch diffs
- uses: ecodrix/erix_buddy@v1
with:
provider: gemini
api-key: ${{ secrets.GEMINI_API_KEY }}
mode: branch
branch: origin/${{ github.base_ref }}
severity: minor
fail-on-blocking: "true"
post-pr-comment: "true"The full input list is in action.yml. Other recipes live in examples/.
| Provider | provider: id |
Required secret(s) | Local? |
|---|---|---|---|
| Google Gemini API | gemini |
GEMINI_API_KEY |
β |
| Google Vertex AI | vertex |
GOOGLE_CLOUD_PROJECT, CLOUD_ML_REGION |
β |
| OpenAI | openai |
OPENAI_API_KEY |
β |
| Anthropic Claude | anthropic |
ANTHROPIC_API_KEY |
β |
| Ollama | ollama |
OLLAMA_HOST (default http://127.0.0.1:11434) |
β |
| OpenRouter | openrouter |
OPENROUTER_API_KEY |
β |
| Groq | groq |
GROQ_API_KEY |
β |
| Azure OpenAI | azure-openai |
AZURE_OPENAI_API_KEY, AZURE_OPENAI_ENDPOINT, AZURE_OPENAI_DEPLOYMENT |
β |
| DeepSeek | deepseek |
DEEPSEEK_API_KEY |
β |
| Mistral AI | mistral |
MISTRAL_API_KEY |
β |
| xAI Grok | xai |
XAI_API_KEY |
β |
| ECODrix Hosted | ecodrix |
ECODRIX_REVIEWER_TOKEN |
β |
The
ecodrixprovider is the optional ECODrix-hosted relay for users who don't want to bring their own key. The reviewer itself never contacts ECODrix unless you explicitly select it.
# Interactive setup (saves credentials to ~/.config/erix-reviewer.json, mode 0600)
npx @ecodrix/erix-buddy
# Review staged changes
npx @ecodrix/erix-buddy --staged
# Audit the whole codebase, only major+ issues
npx @ecodrix/erix-buddy --full --severity major --max-files 200
# Use Ollama, no API key required
npx @ecodrix/erix-buddy --provider ollama --model qwen2.5-coder:7bpnpm dlx and bunx work the same. Full local CLI guide β docs/local-cli.md.
- π Getting started
- π§ͺ GitHub Actions
- π Providers in depth
- βοΈ Configuration reference
- π³ CI recipes
- π» Local CLI
- π©Ή Troubleshooting
- β FAQ
- PR review on every pull request
- Nightly full-codebase audit
- Multi-provider matrix
- Monorepo with path filters
- Self-hosted Ollama runner
- GitLab CI integration
- API keys live only in GitHub Secrets (CI) or
~/.config/erix-reviewer.jsonmode0600(local). They are never sent anywhere except the provider you selected. - Erix Buddy never phones home. There is no telemetry, no usage tracking, no analytics.
- Treat LLM-generated patches as suggestions β always review before applying.
- See
SECURITY.mdfor the responsible disclosure policy.
| CodeRabbit | Erix Buddy | |
|---|---|---|
| Open source | β | β MIT |
| Bring your own LLM key | Limited | β 12 providers |
| Run locally without a server | β | β |
| Local Ollama support (free / private) | β | β |
| GitHub Action | β | β |
| GitHub Marketplace | β | β |
| Inline review comments | β | π‘ in progress |
| Sticky PR walkthrough | β | β |
| Mermaid sequence diagrams | β | β |
| Severity tiers + categories | β | β |
| SARIF export | π‘ | β |
| Self-hostable | β | β |
| Cost | $24/dev/mo | $0 + your provider's metering |
We're not pretending to have full parity yet β see the roadmap. But the trade-off is clear: own your tooling, choose your model, pay only for the tokens you actually use.
We welcome issues, PRs, and recipes! Read CONTRIBUTING.md before submitting.