Show HN: shared verified memory for AI agents: one learns, all recall (kage-core.com)

Kage, a new open-source tool for coding agents, introduces a verified memory system that enables teams to share and learn from each other's experiences seamlessly. Every lesson learned by an agent is transformed into version-controlled files within a team's repository, ensuring that all citations are checked against the codebase to prevent misinformation. This innovative approach allows coding agents to retain memory without risking stale or hallucinated citations, streamlining the development process. The significance of Kage lies in its ability to not only capture memories but also to verify and maintain trust in those memories over time. Each session starts with a digest of previous learnings, and any code changes that could invalidate shared memories are flagged before they are merged. This real-time verification approach enhances collaboration and reduces the likelihood of errors stemming from outdated or incorrect information. Kage's architecture, which requires no accounts or API keys, operates entirely on the user's local environment and syncs over a private git remote, catering to teams looking for a secure and efficient memory management solution in their AI-driven coding workflows.