openintenet 0.1.1

In-memory virtual base for multi-agent intelligence — ephemeral, deterministic, thread-safe, zero external dependencies

OpenInternet

In-memory virtual base for multi-agent intelligence. Zero-dependency, ephemeral, deterministic, and thread-safe.

---

Features

• VirtualBase — A self-contained, in-memory platform that manages agents, knowledge objects, graph relationships, and message routing
• VirtualAgent — Deterministic agents with SHA256-based DIDs, seed-based signing, task handlers, and a fluent async API (publish_many, async_ask, async_publish, async_query)
• Agent Discovery — Find agents by capability, name prefix, or tag using ScalableAgentRegistry with inverted indexes, a trie, and a Bloom filter
• Vector Knowledge Store — 256-partition PartitionedVectorIndex with pre-normalized embeddings, metadata filters, and inverted index acceleration
• Graph Relationships — Thread-safe AdjacencyMap with BFS traversal and evidence chain following
• HyperLogLog — Probabilistic cardinality estimation with LinearCounting correction for sparse registers
• Message Protocol — Rich messages with request_id, error_code enum, timeout_seconds, timestamp, and reply_to
• Snapshot/Restore — Full serialization and deserialization of base state
• Thread Safety — RLock on all stores and the base itself
• Zero External Dependencies — Pure Python, no Docker, no databases, no network services

Quick Start

from openintenet import VirtualBase

base = VirtualBase("MyNetwork")
alice = base.create_agent("Alice", ["research", "qa"])
bob = base.create_agent("Bob", ["translation"])

alice.publish("Climate data shows warming trends.", tags=["climate"])
results = alice.query("climate")
reply = alice.ask(bob, "Translate 'hello' to French")

agents = alice.discover_agents(capability="research")

License

MIT