Deploying and managing production-grade etcd clusters on cloud providers: failure recovery, disaster recovery, backups and resizing.
Distributed Saga transaction orchestrator executing forward actions and compensating reverse rollbacks
Distributed Saga transaction orchestrator executing forward actions and compensating reverse rollbacks
Production style autonomous agent framework in Python + React that plans multistep tasks, persists state with SQLite, and recovers from failures using exponential backoff retry logic. Includes real time monitoring dashboard, deterministic recovery demo, and API for task orchestration.
A programmatic autonomous LLM agent
🎛️🔁🚀 A tiny, zero‑dependencies retry helper with exponential backoff + jitter—usable for KV, HTTP, Durable Objects, or any async function.
Implementation of a distributed database with multi-version concurrency control, replication, deadlock detection and failurerecovery.
Semantic failure memory for AI agents — record reasoning failures, search past failures before acting, stop rediscovering the same mistakes.
Failure-aware Python runtime for tool-using LLM workflows with bounded recovery, output verification, tracing, and reproducible evaluation.
Evidence-checked survey and data atlas of automatic reset, failure recovery, safety, and sustained autonomy in real-world robot RL.
Distributed RAG ingestion pipeline with idempotent vector indexing, DLQ-backed retries, and deterministic failure recovery for production AI systems
Built a 6-agent Python pipeline where agents detect their own reasoning failures, classify them, retrieve similar past failures from memory using RAG, and recover from the exact step where things went wrong. Uses a Bayesian estimator to decide whether recovery is worth retrying. Powered by Gemini 2.5 Flash.
证据约束的开源研究 Agent：只读 GitHub 适配器、计划修订、严格取消与引用审查。
An experimental agentic overlay for Zelda: Breath of the Wild, combining screenshots, speech, persistent campaign memory, and a comprehensive knowledge base to spot overlooked details, explain the world, and provide context-aware guidance as you play.
专为 OpenClaw 打造的 Browser Operations Platform：intelligence、action policy、handoff、autopilot、failure recovery 全进内核
Hierarchal MultiModel AI Agentic Orchestration Platform - Project J.A.R.V.I.S ★ Echo Mac is engineered to be a state-of-the-art, fully autonomous agentic wrapper for your operating system. It operates at the intersection of streaming multimodal AI and deep OS integration.
证据驱动的 Browser Agent 运行时：Plan–Act–Verify–Recover、有界恢复与确定性 Playwright 演示。
PostgreSQL paper-execution boundary: durable state, retries, concurrency, and fresh authorization. No broker. No real-money path.
Research protocol for grasp recovery under stale observations, intervention delay, and hidden contact dynamics.
History-conditioned π0.5 failure recovery on LIBERO with controlled-slip evaluation and recovery-trajectory distillation.
To associate your repository with the failure-recovery topic, visit your repo's landing page and select "manage topics."