arthur

Inspired by Quantum Merlin-Arthur (QMA) verification, Arthur is a deterministic ground truth verification method for AI-generated code. It catches hallucinated file paths, schema references, imports, env vars, routes, and package API usage before code gets written.

AI coding assistants hallucinate. They reference files that don't exist, query database columns that aren't real, import packages that aren't installed, and build plans on assumptions that don't match the actual codebase.

The model can't reliably catch its own mistakes. Self-review is limited by the same context constraints and attention budget that caused the hallucinations in the first place.

Arthur runs deterministic checks against ground truth (your actual files, schemas, packages, env vars) and returns the results, including what actually exists, so the model can self-correct.

claude mcp add arthur -- npx arthur-mcp

Arthur is now available as an MCP server in Claude Code. All tools run locally. No API key, no credits, no config.

1. Claude Code generates a plan
2. Claude Code calls Arthur's check_all tool
3. Arthur validates every reference against ground truth (file tree, schemas, node_modules, .env files, routes)
4. Arthur returns findings with the correct values: not just "this is wrong" but "this is wrong, here's what actually exists"
5. Claude Code reads the findings and corrects its plan

# When Arthur finds a hallucinated Prisma model:
✗ prisma.engagement - hallucinated-model -> prisma.participantEngagement
  Available models: participant (Participant), contentItem (ContentItem),
                    participantEngagement (ParticipantEngagement)

# When Arthur finds a hallucinated file path:
✗ src/models/User.ts - NOT FOUND
  Closest: src/lib/db.ts, src/app/api/participants/route.ts

# When Arthur finds a wrong Supabase column:
✗ .select('...early_decision_deadline...') - hallucinated-column
  Available columns on