Agent Optimization Hierarchy: Architecture Over Human Review
MikeBirdTech · x · 2026-08-17
Citing @poteto, this post argues that when correcting an agent, one should aim to eliminate the need for intervention entirely. The proposed hierarchy of value is:
- Categorically eliminate the problem via better architecture/data structures
- Turn it into a lint rule or test for CI
- Turn it into a skill or rule
- Human review (discouraged)
The approach emphasizes combining statically typed codebases with deterministic checks and asking the agent how to prevent issues upstream.
More from coding & agent
- Mediocre Agent Data Science Output? Needs Custom Skills — daveholtz · 2026-08-17
- AI Agents Fail to Respect Field-Specific Norms in Academic Research — daveholtz · 2026-08-17
- OKF vs wikillm: How to choose for agent context? — frakc · 2026-08-17
- Bad code means bad agents: Software fundamentals matter more than ever — mattpocockuk · 2026-08-17
- How to gate Agent actions in production environments? — Excellent-Park-1160 · 2026-08-17
- Open Source UnFlow: Graph-based ML Experimentation Tool — ha2emnomer · 2026-08-17