Open Code Review MCP Server

Directory scans and Git branch comparisons bring Open Code Review into AI-assisted development and pull-request review. The scan_directory tool looks for invented imports, nonexistent packages, outdated APIs, and security anti-patterns across source files. Its language coverage includes Python, Go, Java, Kotlin, TypeScript, and JavaScript. For reviews focused on a proposed change, scan_diff examines differences between branches and restricts analysis to modified files and lines, giving maintainers a targeted workflow for checking generated contributions.

Findings can be investigated with explain_issue, which adds category context and guidance on corrections. The heal_code tool retrieves a file and assembles repair context for an AI agent to use; applying the fix is the agent's responsibility. Beyond MCP, the project provides a CLI and integrations for GitHub Actions and GitLab CI. Its deeper analysis supports local Ollama models or OpenAI-compatible providers, and reports can be produced as terminal output, JSON, SARIF, or HTML.