Graph Engineering: The Karpathy Loop, Improved 1000x by Itself — The Anthropic Playbook

This independently compiled July 2026 technical note connects Andrej Karpathy’s autoresearch and AgentHub sketches with Anthropic workflow and knowledge-graph patterns to propose an incremental architecture for verifiable, collaborative, provenance-aware agent systems. The source title’s ‘1000x’ is rhetorical, not a measured improvement claim; the note separately discusses workflows that can schedule up to 1,000 agents.

4 sections · 12 lessons

Course outline

Measured Loops

  1. Externalize the Bottleneck
  2. Program the Program
  3. Build a Safe Ratchet
  4. Bound an Extraction Loop

Collaborative Search

  1. Preserve Experiment Lineage
  2. Coordinate Without Transcripts
  3. Generate and Reduce Work
  4. Design a Migration Workflow

Durable Graphs

  1. Extract Typed Claims
  2. Resolve Without Corruption
  3. Query Bounded Evidence Paths
  4. Investigate an Incident Path

Production Discipline

  1. Join Lineage and Provenance
  2. Separate the Five Planes
  3. Budget, Stop, and Evaluate
  4. Approve a Graph-Grounded Swarm

Start learning with Wondering