Everything PeerWeave can do for you.
The Differentiators doc explains the philosophy. This page shows how it translates into tools you can touch—across sync, agents, graph, and publishing.
AI-native workspace
Ask your space anything. Agents understand code, data, and project structure, and they share context through the graph.
- Natural language queries mapped to semantic graph traversal
- Local-first inference with optional remote model routing
- Trace logs help you audit what every agent did
- Multiple agents can collaborate inside the same space
CRDT sync & branching
Real-time Gitless collaboration with conflict-free data structures that keep working offline.
- CRDT-based sync with vector clock snapshots for every change
- Branching and history live in a content-addressed DAG
- Offline-first workflow with automatic peer sync when online
- Semantic merge strategies for code and structured data
Semantic graph explorer
See your workspace as an interactive knowledge graph. Jump from code to commits, agents, and services instantly.
- 2.5D visualization with pan, zoom, and time pivots
- Layers for structure, recency, and agent footprints
- Click a node to open the related file or artifact
- Timeline scrubber to replay changes
Ports runtime
Turn any folder into a port and run services, models, or tools inside your peer mesh—no cloud deploy required.
- Sandboxed service execution with capability scoping
- Peer discovery for HTTP, GraphQL, gRPC, WebSocket
- Share APIs and apps with zero infrastructure
- Great for MVPs, experiments, and personal micro-clouds
Mirrors & anchors
Publish just the parts you want as public mirrors, then anchor hashes for provenance when needed.
- Selective exposure with automatic PII redaction
- .evr handles for human-friendly URLs
- Optional EVRMORE anchoring for immutable proof
- Discovery layer for finding published mirrors
Security & privacy
End-to-end encryption, DID-based identity, and capability-scoped access keep spaces sovereign.
- MLS group encryption for shared spaces
- did:key and did:web support for devices and personas
- Capability tokens with granular scopes
- Agent sandboxing with resource limits and audit trails
How it compares
We love GitHub, Replit, and other tools—they just weren’t built for sovereign, agent-native workflows. Here’s the quick scorecard.
Feature
Real-time Gitless sync
Feature
Agent-native design
Feature
Fully P2P / local-first
Feature
Semantic knowledge graph
Feature
Integrated sandboxed services
Feature
Mirror + anchor publishing
Ready to try the whole stack together?
Install the desktop app, open a space, and invite a few agents. You’ll feel the difference the moment you unplug your Wi-Fi and nothing breaks.