v0.1.0: ReLoop — Failure memory for AI agents

ReLoop v0.1.0

Every agent fails. ReLoop is the first framework that gets smarter from failure.

What's included

• REJD Loop — Retrieve, Execute, Judge, Distill self-healing cycle
• OpenAI Agents SDK orchestration with 4 specialist agents and handoffs
• Redis failure memory — structured records with semantic search, vector embeddings, confidence decay
• Context Surfaces — expose failure memory as auto-generated MCP tools
• Timeline UI — RED to YELLOW to GREEN visual execution timeline
• Blaxel sandbox — Firecracker microVM with 25ms checkpoint/restore
• A/B comparison mode — with memory vs without memory side-by-side

Three ways to use

As a library (3 lines):

from reloop import FailureMemory
memory = FailureMemory(redis_url='redis://localhost:6379')
similar = await memory.search('ImportError sharp')

As a framework:

pip install reloop
reloop init && reloop demo

As an MCP server:
Add to your Claude Code / Cursor config.

Dashboard

Live at: https://dashboard-five-pi-36.vercel.app

Stats

• 182 tests passing
• 10 API endpoints + 10 Context Surfaces endpoints
• Integrations: OpenAI Agents SDK, LangGraph, CrewAI, Claude Agent SDK
• Apache 2.0 license