58 / 108 · 03 Agentic Testing Architectures · Case Study: The OpenObserve Council of Sub-Agents← prev⊞ allnext →☰ Read as one page
7.6Applying OpenObserve's Lessons to Your Projects
| OpenObserve Practice | How to Apply It | Minimum Team Size |
|---|---|---|
| Specialized agents | Start with 3: Analyzer, Generator, Runner | 1 engineer |
| Shared state file | Use a JSON file or SQLite database | 1 engineer |
| Human approval gates | Require PR review for agent-generated tests | 1 engineer |
| Incremental execution | Only analyze changed files since last run | 1 engineer |
| Flaky detection | Run tests 3-5x before merging | 1 engineer |
| Token budgets | Set per-module limits in config | 1 engineer |