Skip to content

MemFuse v0.1.0

Latest
Latest

Choose a tag to compare

View all tags
@M1n9X M1n9X released this 01 Jun 15:53
· 14 commits to main since this release
v0.1.0
cda9aa0

MemFuse v0.1.0 — Initial Release

Persistent memory hub for AI coding agents.

Server (this release)

  • Single binary — Rust/Axum HTTP server with embedded SQLite, zero external dependencies
  • 117 REST API endpoints with OpenAPI/Swagger docs
  • Memory pipeline: sessions → observations → episodes → facts → consolidation
  • Multi-strategy search: precision, diverse, recent, comprehensive (LLM-assisted)
  • Tiered semantic index: L0 abstract / L1 overview / L2 full content
  • Code import: localfs, git, URL connectors with AST skeleton extraction (Rust, Python, TS, JS, Go, Java, C/C++)
  • Heuristic learning: draft → candidate → confirmed lifecycle
  • Service management: systemd, macOS LaunchAgent, Docker support
  • Security: API key auth, CORS, SSRF protection, rate limiting

SDK (npm)

Published separately as @percena/memfuse:

  • CLI with 110 commands (100% API coverage)
  • MCP server with 43 agent-facing tools
  • 8 lifecycle hooks for Claude Code / Codex
  • Type-safe HTTP client library
  • Skill loader for agent platforms

Quick Start

# Download and extract
tar -xzf memfuse-server-0.1.0-darwin-arm64.tar.gz
./memfuse-server

# Or build from source
cargo build --release -p mfs-server
./target/release/mfs-server

# Verify
curl http://127.0.0.1:8720/health

# Install SDK
npm install @percena/memfuse
npx memfuse setup install --platform=claude-code

Platforms

Asset Platform
memfuse-server-0.1.0-darwin-arm64.tar.gz macOS Apple Silicon

Linux x86_64 binary will be available in the next release via CI cross-compilation. For now, build from source: cargo build --release -p mfs-server

Assets 3
Loading