Skip to content

~/courses $ cd c1

Agentic Architecture & Orchestration

C1 · 27% of the exam · 6 lessons · about 58 min of reading

How agents run: the loop and its stop conditions, the Agent SDK, orchestrators and parallel subagents, handoffs, human approval, and when an agent is the wrong tool.

Course progress0/6

Short by the site author (n_tech_am) on YouTube, in Armenian. Pressing play loads it from youtube-nocookie.com. ·open on YouTube

  1. The agentic loop and stop conditionsHow a tool-using agent cycles between the model and your code, and every way that loop is supposed to end.10 min
  2. Agent SDK basicsEmbed Claude Code's agent loop in your own Python or TypeScript program, with limits, permissions and a result you can check.10 min
  3. Orchestrator–worker and parallel subagentsSplit a task across focused subagents that run in parallel, each with its own context and tools, and keep the fan-out bounded.10 min
  4. Handoffs and failure isolationWrite briefs a fresh subagent can act on, get structured reports back, and keep one worker's failure from sinking the whole run.9 min
  5. Human-in-the-loop approvalsPut a person in front of the calls that matter, using the permission order, canUseTool, hooks and plan mode.10 min
  6. When not to use agentsWorkflows, single calls and plain code often beat an agent. How to choose the simplest design that works.9 min