Worked on the mastra-ai/mastra repository to resolve a critical deadlock issue in the Observational Memory system. Addressed the problem by deriving unique internal thread IDs for observer and reflector agents, preventing circular waits and enhancing stability during nested agent executions. The solution involved isolating thread logic within the @mastra/memory package, removing special-case handling at the core level, and updating observer and reflector streams to use the new context. Regression tests and updated payload tests were introduced to ensure reliability. Utilized TypeScript, Node.js, and full stack development skills to improve multi-agent workflow reliability and reduce production deadlocks.