OpenViking
Self-evolving context database giving agents organized, searchable memory
Agent memory that learns instead of just recalling
Hindsight is a memory system for AI agents that distinguishes retrieval from actual learning, organising memory into world facts, experiences, consolidated observations and synthesized mental models. It targets teams building support, research or project-management agents that must stay coherent across months of sessions.
Memory is structured in four types and driven by three operations: retain pushes memories in, recall runs four retrieval strategies in parallel and fuses the results, and reflect forms new connections between memories. A one-command installer wires it into detected coding agents and builds a per-repository memory bank from git history and past sessions; self-hosting is a single docker run, with a hosted option available.
Top 10 Fastest Growing Repositories on GitHub Today - Top 10 GitHub Repos (2026)
Top 10 Github Repos That Blew Up This Week (Oct 4, 2026)
Free on GitHub: manage agents, 11Labs alternative, incredible design & more
Self-evolving context database giving agents organized, searchable memory
Fully local voice cloning, dubbing and transcription studio
Compress tool outputs, logs and RAG chunks before the LLM sees them
Graph-native company brain giving agents structured, accountable context
Visual web search for agents using screenshots, not just text
Self-hosted net-worth tracker connected to thousands of banks