Become an AI engineer

# Agentic harness

Agentic harness is a unit of 10 lessons in Rabbithole's Become an AI engineer course, at levels 3 to 4.

## Lessons

1. Human in the loop: pausing a workflow for a person · Level 3
2. Agent Deployment: From Local Agent to Running Service · Level 3
3. The agent loop · Level 3
4. Agent SDKs and frameworks: what they run for you · Level 3
5. Guardrails: checks around an LLM call · Level 3
6. Secrets Management: Keep Credentials Out of Your Code · Level 3
7. Designing tools for agents · Level 3
8. Tool Search: Let the Model Look Up Its Tools · Level 3
9. Why agents fail, and how to test for it · Level 4
10. Harness engineering: the system around the model · Level 4

## Sources

These lessons draw on 19 sources. The most used:

- [Agent Development Kit (ADK)](https://adk.dev/)
- [OpenAI Agents SDK](https://openai.github.io/openai-agents-python/)
- [SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineering](https://arxiv.org/abs/2405.15793)
- [AgentBench: Evaluating LLMs as Agents](https://arxiv.org/abs/2308.03688)
- [Claude Cookbooks](https://github.com/anthropics/claude-cookbooks)
- [OpenAI Guardrails Python](https://openai.github.io/openai-guardrails-python/)
- [Claude Docs — Tool Use and Agent Tools](https://platform.claude.com/docs/en/agents-and-tools/tool-use/overview)
- [Codex as a platform: build on the open agent harness](https://developers.openai.com/blog/codex-as-a-platform)

[All units in Become an AI engineer](https://rabbithole.pro/courses/ai-engineer)

Source page: https://rabbithole.pro/courses/ai-engineer/agentic-harness