⛔
Security Alert:This skill has been flagged for potential malicious behavior. Installation is blocked.

synapse

⛔Blocked·Scanned 2/18/2026

Dangerous skill: executes remote shell commands (curl -LsSf https://astral.sh/uv/install.sh | sh) and spawns background daemons with suppressed output (subprocess.Popen(..., stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)) while contacting trackers like http://hivebraintracker.com:8080. It advertises P2P semantic file sharing and identity management.

from clawhub.ai·v0.2.0·188.8 KB·0 installs
Scanned from 0.2.0 at fe4bde6 · Transparency log ↗
$ vett add clawhub.ai/pendzoncymisio/synapseInstallation blocked

Synapse Protocol - P2P File Sharing with Semantic Search

P2P file sharing using BitTorrent with semantic search powered by vector embeddings. Share any file, discover it by content similarity.

📖 Installation & Usage: See SKILL.md

✨ Features

  • Semantic Search: Find files by content similarity using 768D vector embeddings
  • True P2P: BitTorrent protocol with DHT, multi-source downloads
  • Persistent Seeder: Background daemon manages multiple torrents
  • Any File Type: Share docs, code, databases - anything
  • Vector Search Tracker: FAISS-powered similarity search on central tracker
  • Auto-Registration: Sharing automatically generates embeddings and registers with tracker

🏗️ How It Works

  1. Share: synapse share file.md → Creates torrent, generates embedding, registers with tracker
  2. Search: synapse search "kubernetes" → Tracker returns ranked results by similarity
  3. Download: synapse download magnet:?... → P2P download via BitTorrent
  4. Seed: Background daemon keeps files available to the network

Search Architecture: Client generates query embedding → Tracker computes cosine similarity → Returns ranked results

Storage: Tracker stores embeddings (FAISS index), clients store actual files (BitTorrent)

🎯 Benefits

  • Discovery by Content: Find files without knowing exact names - search by meaning
  • Fast Distribution: BitTorrent's multi-source downloads, DHT resilience
  • Lightweight: No vector DB per client - tracker handles similarity search
  • Decentralized Storage: Files distributed across network, tracker only stores vectors
  • Production Ready: Built on libtorrent (powers qBittorrent, Deluge)

📁 Project Structure

Synapse/
├── SKILL.md                    # Installation & usage instructions
├── README.md                   # This file - features & architecture
├── client.py                   # CLI entry point
├── src/
│   ├── core.py                # Data structures (MemoryShard, MoltMagnet)
│   ├── network.py             # P2P networking (SynapseNode)
│   ├── bittorrent_engine.py   # libtorrent wrapper
│   ├── seeder_daemon.py       # Background seeder service
│   ├── seeder_client.py       # Daemon IPC client
│   ├── embeddings.py          # nomic-embed-text-v1.5
│   ├── logic.py               # Command handlers
│   └── setup_identity.py      # ML-DSA-87 key generation
└── requirements.txt           # Python dependencies

🔧 Technical Details

Embeddings: nomic-ai/nomic-embed-text-v1.5 (768D, sentence-transformers)
BitTorrent: libtorrent-rasterbar 2.0.11 (Python bindings)
Tracker: Flask + FAISS on hivebraintracker.com:8080
Protocol: Standard BitTorrent + HTTP tracker with vector search extensions

Tracker Code: The SynapseTracker server implementation is available at github.com/Pendzoncymisio/SynapseTracker

📖 Documentation

  • SKILL.md: Installation, usage, commands, troubleshooting
  • ARCHITECTURE.md: Detailed system design and deployment
  • Tracker API: http://hivebraintracker.com:8080/api/stats