Show HN: Sibyl – self-hosted cross-agent memory for AI coding agents (github.com)

Sibyl, a recently launched self-hosted cross-agent memory system for AI coding tools, allows users to maintain a unified knowledge graph that retains and shares context across different coding agents like Claude Code and Codex. This transforms the way AI tools handle user memory, ensuring that data stays with the user rather than being controlled by third-party services. The CLI-centric interface requires no SDKs and supports various deployment options, enabling seamless integration across different environments. The significance of Sibyl lies in its innovative approach to user memory, which enhances AI productivity by compounding context over time. Users can benefit from features like semantic search that allows nuanced retrievals based on meaning, and a task workflow that integrates with various agents for project management. The architecture promises not only data security and privacy through its self-hosting model but also an efficient memory loop that assists both users and AI agents in learning and recalling information. This empowers developers and teams to build and debug more effectively while creating a personal knowledge repository less susceptible to vendor lock-in.