Docs/Concepts

Concepts

Understand the architecture and core ideas that make Seshat different.

01Available

What is Seshat?

The runtime, the platform layer, and the thesis behind organized agentic work.

Read
02Available

Mono-Run Architecture

How one runtime owns sessions, prompts, tools, permissions, and persistence.

Read
03Available

Runtime Modes

Entry surfaces like CLI, gRPC, HTTP, and SDK versus execution modes like execute, plan, and pair_programming.

Read
04Available

Agent Teams and Mailboxes

Persistent agent identities, mailbox transport, dispatcher routing, and the beginning of a real team runtime.

Read
05Available

Memory and RAG

Session persistence, compaction, durable memory, knowledge-graph memory tools, and retrieval-augmented generation.

Read
06Available

Skills and MCP

How reusable skills shape behavior and how MCP servers extend the runtime with external tools, resources, and prompts.

Read
07Available

Tools & Providers

How the tool surface and provider layer fit together across built-ins, external capabilities, and model providers.

Read