​Building CLI / Coding Agents that Learn

​Join us for the 7th installment of the Stateful Agents meetup series, in-person in San Francisco! 🌉 👾

​CLI / terminal agents like Claude Code, Gemini CLI, and OpenAI Codex are perhaps the closest we've gotten to real, autonomous agents that can achieve long-running tasks independently. Yet they still suffer from the fundamental limitation of agents today: they are stateless, and have limited long-term context management (think /compact in Claude Code).

​For this month's Stateful Agents Meetup, we'll walk through how to build your own CLI agent harness that is:

• ​Model-agnostic (can work with major providers like Claude and GPT-5, as well as frontier open weights models like GLM-4.6, M2, and K2)

• ​Persistent (messages are persisted and searchable)

• ​Built-in context management (avoid compaction and enable long-term learning)

• ​Configurability (add fun custom personalities)

​What to expect

​This meetup will be a live workshop - we'll be building and demoing frontier coding agents live. Come by to learn about:

• ​How to build your own open CLI agent harness on Letta's API

• ​How stateful CLI agents can use memory or skills to learn and improve over time

• ​How to architect an agent's memory structure to work well with coding - what should coding agents remember over time, and what should they forget?

​Agenda

• ​5:30pm: Doors open

• ​6:00-7:15pm: Workshops and demonstrations

• ​7:15-8:30pm: Food and networking