| # Resources | |
| ## Primary Sources | |
| 1. **Samuel's META2 Antworten** (local, highest trust) | |
| - `frage_05_qualitaet/ANTWORT.md` — Quality metrics for agent work | |
| - `frage_07_ueberforderung/ANTWORT.md` — Overload and parallel system breaks | |
| - `frage_01_kontrolle/ANTWORT.md` — Control and meta-management | |
| - `frage_04_thalamus/ANTWORT.md` — Digital thalamus / filtering | |
| 2. **Hermes Agent Documentation** (local) | |
| - `~/TH-Mannheim/` and `~/Projekte/Agent-Baukasten/` — existing agent architectures | |
| - Hermes skills and registry system | |
| ## External Sources (to evaluate) | |
| - [ ] Andy Grove, "High Output Management" — output-focused management, OKRs | |
| - [ ] Kanban/WIP limits literature — vertical depth vs horizontal breadth | |
| - [ ] Software craftsmanship: "Definition of Done" patterns | |
| - [ ] Agent evals: finding reliable quality metrics for LLM pipelines | |
| ## Trust Levels | |
| - **High**: Samuel's own META2 analysis (self-generated, self-verified) | |
| - **Medium**: Hermes architecture docs (designed but not all operationalized) | |
| - **Low**: General productivity/management literature (needs filtering through Samuel's actual context) |