Vault
Provides local-first memory governance for AI agents, enabling shared, reviewable, and auditable memory via SQLite and MCP.
Provides local-first memory governance for AI agents, enabling shared, reviewable, and auditable memory via SQLite and MCP.
Vault Agent Memory offers a governed, shared memory layer designed for AI agents like Codex, Claude Code, and Hermes. Unlike traditional notes apps or vector databases, it focuses on the lifecycle of agent memory, addressing common failures such as agents acting like new sessions, old notes overriding new decisions, and sensitive information leaks. It empowers teams to govern what agents remember, trust, share, forget, and roll back, ensuring memory is auditable, current, and relevant through a structured "propose -> review -> promote -> search -> bounded read -> rollback -> audit" workflow.