Skip to content

QUICKSTART.md

Latest commit

 

History

History
81 lines (60 loc) · 1.68 KB

QUICKSTART.md

File metadata and controls

81 lines (60 loc) · 1.68 KB

Quickstart Guide

Get PR-Agent running with local Ollama in 2 minutes.

Prerequisites

  • Docker or Podman
  • Ollama instance with a model (e.g., codestral:22b)
  • Git provider access token (Gitea, GitHub, or GitLab)

1. Pull the Image

docker pull ghcr.io/tobend/pr-agent:latest

2. Run a Review

Replace the placeholder values:

docker run --rm \
  -e CONFIG.GIT_PROVIDER=gitea \
  -e GITEA.URL=https://your-gitea.com \
  -e GITEA.PERSONAL_ACCESS_TOKEN=your-token \
  -e OPENAI.API_BASE=https://your-ollama.com/v1 \
  -e OPENAI.KEY=dummy \
  -e CONFIG.MODEL=openai/codestral:22b \
  ghcr.io/tobend/pr-agent:latest \
  --pr_url="https://your-gitea.com/owner/repo/pulls/1" review

Available Commands

Command Description
review Full code review with issues and score
describe Generate PR description
improve Suggest code improvements

Provider Configuration

Gitea:

-e CONFIG.GIT_PROVIDER=gitea
-e GITEA.URL=https://...
-e GITEA.PERSONAL_ACCESS_TOKEN=...

GitHub:

-e CONFIG.GIT_PROVIDER=github
-e GITHUB.USER_TOKEN=...

GitLab:

-e CONFIG.GIT_PROVIDER=gitlab
-e GITLAB.URL=https://...
-e GITLAB.PERSONAL_ACCESS_TOKEN=...

Recommended Models

  • codestral:22b - Best for code review
  • qwen2.5-coder:14b - Good alternative
  • mistral:latest - Lightweight

Troubleshooting

"Failed to parse review data"

  • Ensure Ollama is v0.5+ (supports json_schema)
  • Check model is loaded: curl your-ollama:11434/api/tags

Connection refused

  • Verify Ollama is accessible from Docker
  • Check firewall/network settings

Next Steps

See README.md for full documentation.