LLM Coding Agents

A Phase Shift in Software Engineering (Late 2025)

Core Observations

Failure Modes & Limits

Subtle Errors

Overengineering

Models will happily write 1,000 lines of brittle code until you ask, “Couldn’t this be 100?”

Operational Risks

How to Get Leverage from Agents

Prompting Strategy

Engineering Tactics

Don’t tell agents how to do the work. Tell them what “done” looks like.

Productivity Effects

Speedup vs. Expansion

Stamina & Flow

The bottleneck moves from effort to judgment.

Risks & Long-Term Implications

Skill Atrophy

The “Slopacolypse”

Open Questions

Where This Leaves Us

Key Takeaway:
Coding agents are already a net productivity win, but they demand constant human verification. The real leverage comes from shifting from “writing code” to “defining outcomes and verifying results.”