Merge pull request #8 from CodeSignal-Learn/dev

Improve polling cadence and avoid overlapping state requests

Merge pull request #7 from CodeSignal-Learn/dev

Handle run-loop failures in agent

Blueprint and agent.py up to date with content

Match blueprint from engine

Updating blueprint (course 3) to match app

Matching bluprint Course 2 unit 4 agent loop to app

Updating blueprint (course 2) to match final app

Include blueprint to version control

Creating agents.md

Update documentation

Serializing LLM context as a single user message

Using required openai parameter instead of prompting for tool use

Hardcoding schemas and flattening the match statement