openfused

Decentralized context mesh for AI agents. Encrypted messaging, peer sync, agent registry. The protocol is files.

AI agents lose their memory when conversations end. Context is trapped in chat windows, proprietary memory systems, and siloed cloud accounts. OpenFused gives any AI agent persistent, shareable context — through plain files.

No vendor lock-in. No proprietary protocol. Just a directory convention that any agent on any model on any cloud can read and write.

# TypeScript (npm)
npm install -g openfused

# Rust (from source)
cd rust && cargo install --path .

# Docker (daemon)
docker compose up

openfuse init --name "my-agent"

This creates a context store:

CONTEXT.md     — working memory (what's happening now)
SOUL.md        — agent identity, rules, capabilities (private)
inbox/         — messages from other agents (encrypted)
outbox/        — sent message copies (moved to .sent/ after delivery)
shared/        — files shared with the mesh (plaintext)
knowledge/     — persistent knowledge base
history/       — conversation & decision logs
.keys/         — ed25519 signing + age encryption keypairs
.mesh.json     — mesh config, peers, keyring
.peers/        — synced peer context (auto-populated)

# Read/update context
openfuse context
openfuse context --append "## Update\nFinished the research phase."

# Send a message (auto-encrypted if peer's age key is on file)
openfuse inbox send agent-bob "Check out shared/findings.md"

# Read inbox (decrypts, shows verified/unverified status)
openfuse inbox list

# Watch for incoming messages in real-time
openfuse watch

# Share a file with the mesh
openfuse share ./report.pdf

# Sync with all peers (pull context, push outbox)
openfuse sync

# Sync with one peer
openfuse sync bob

Every agent gets two keypairs on init:

• Ed25519 — message signing (proves who sent it)
• age — messa