What is quorum?

Quorum is a native macOS app that turns a single question into a small team of AI research agents working in parallel, then merges their findings into one cited note. It sits on top of the Claude Code command line tool, acting as the orchestrator and supervisor while the agents themselves do the actual searching. Each agent only gets read-only tools like web search and web page fetching, so even if a run goes wrong it cannot change any of your files, it can only waste money. When you ask a question, a planner first splits it into several different angles to investigate, and you can review and edit that plan before anything runs. Each angle is handled by its own agent, blind to what the other agents are doing, and a supervisor watches spend and time so a run cannot run away. Once all the agents finish, a summarizer combines their findings into a single note, checks that every claim has a real citation, and flags any contradictions or gaps. If something is still unresolved, another round of research can run just on those pieces. All of your notes are stored as plain markdown files with wikilinks in a project folder you choose, so they work with git, plain text search, or apps like Obsidian and Logseq, with no database or cloud storage involved. A separate lint pass can scan your whole collection of notes for contradictions and missing links. You can pick different cost presets and set spend limits per topic or per run, and because Quorum reuses your existing Claude Code sign in, there is no separate API key to manage. The project includes a small benchmark comparing Quorum against plain Claude Code and ChatGPT Deep Research on four research questions.