Glossary

Key terms and concepts used throughout kernl.

1.1 :: Agent. A language model (LLM) which is configured with instructions, model settings, and tools. Agents are stateless — they define behavior but don’t hold conversation state.

1.2 :: Thread. A stateful execution of an agent with a clear lifecycle, which consists of a sequence of events.
    (a) State. Temporary data relevant only during the current session.
    (b) Event. The content of the conversation, as well as actions taken by agents such as tool calls.

1.3 :: Memory. Knowledge that persists beyond individual threads. Can be created, searched, and updated by agents or manually.

1.4 :: Context. Runtime data passed into an agent’s execution — user info, session state, or any data your agent needs access to. It is a propagation mechanism which carries execution-scoped values across API boundaries.

1.5 :: Toolkit. A collection of related tools that can be attached to an agent (e.g. gmail :: {list, send, ...}).
    (a) Tool. A function the LLM can call to take actions or retrieve information.

1.6 :: Namespace. A logical partition for isolating threads and memories. Useful for multi-tenant applications.