Main development is on decentralized git: htree://npub1xdhnr9mrv47kkrn95k6cwecearydeh8e895990n3acntwvmgk2dsdeeycm/nostr-social-graph

A TypeScript library for building and querying social graphs from Nostr follow events.

• Build social graphs from Nostr follow events
• Query followed users, followers, and follow distances
• Change social graph root user with efficient distance recalculation
• Low memory consumption
• Efficient binary serialization (55% smaller than JSON)
• Pre-crawled datasets
• Server for maintaining and serving the up-to-date social graph, for quick initialization in web apps

See tests for detailed usage examples.

• Demo: graph.iris.to (examples dir)
• Documentation: mmalmi.github.io/nostr-social-graph/docs
• API Endpoints:
  • https://graph-api.iris.to/social-graph?maxBytes=2000000
  • https://graph-api.iris.to/profile-data?maxBytes=2000000&noPictures=true
• Used in production at iris.to.

To point the examples search at a hashtree index, set VITE_PROFILE_SEARCH_INDEX=nhash1qqsgm4ex4d4dxgz39hj6q7t7ax7u4k57gp2zkjuxtfga7wpw6dy6xpg9yqu6y09zecw9hzettkaulu928dt58ndt0h2exw6qg5kxyrprucz0cukym2c (and optionally VITE_BLOSSOM_SERVERS=url1,url2). Latest published profile search index (2025-01-23): nhash1qqsgm4ex4d4dxgz39hj6q7t7ax7u4k57gp2zkjuxtfga7wpw6dy6xpg9yqu6y09zecw9hzettkaulu928dt58ndt0h2exw6qg5kxyrprucz0cukym2c. To publish the profile search index to Blossom, run BLOSSOM_NSEC=... yarn publish-profile-index.

The TypeScript implementation lives in SocialGraph.ts, and the Rust workspace lives under rust/.

The Rust workspace now has two interchangeable backends:

• rust/crates/nostr-social-graph: in-memory core graph and binary format
• rust/crates/nostr-social-graph-heed: optional LMDB/heed backend for persistent large graphs

Both Rust backends implement the shared SocialGraphBackend runtime trait from the core crate.