Hi HN — I’m Andrea. I built Bob the Fixer to make “AI fixes” less vibes, more measurable.
The idea: run SonarQube locally (containerized), then expose an MCP server that gives AI coding CLIs a fix-ready view of your static analysis results. Instead of “guess what to fix”, the assistant can:
- scan a repo + apply quality gates
- fetch rich issue details (rule info + plain-text code context)
- prioritize work (tech debt / coverage gaps / duplication / patterns)
- iterate: fix → test → re-scan
Important: this is not a 1:1 “SonarQube API over MCP” wrapper — it’s an opinionated remediation workflow.
The idea: run SonarQube locally (containerized), then expose an MCP server that gives AI coding CLIs a fix-ready view of your static analysis results. Instead of “guess what to fix”, the assistant can:
- scan a repo + apply quality gates - fetch rich issue details (rule info + plain-text code context) - prioritize work (tech debt / coverage gaps / duplication / patterns) - iterate: fix → test → re-scan
Important: this is not a 1:1 “SonarQube API over MCP” wrapper — it’s an opinionated remediation workflow.
Quickstart (Linux/macOS): curl -fsSL https://raw.githubusercontent.com/andrearaponi/bob-the-fixer... | bash
Then open your AI CLI inside a repo and ask something like: “Scan this project with Bob the Fixer, then show me details for the top critical issue.”
Story / why it exists (incl. the “two weeks of tech debt in one night” bet):
https://andrearaponi.it/open-source/bob-the-fixer/