'AI Can Do Everything Except Go to Jail for You': Quality Rules for Agent Coding
dotey · x · 2026-09-19
A long-form essay on reliably shipping code in the age of agent coding. The author opens with the punchline: AI can do everything for you except take the blame—responsibility for AI-generated code still rests with humans.
The piece first outlines the productivity advantages of agent coding, then focuses on how humans add value inside the agent loop: where to intervene, how to review and gate AI output, so that code produced by agent programming can be delivered reliably.
More from coding & agent
- gog CLI ships an MCP server for Google Workspace with narrow, read-only-by-default tooling — steipete · 2026-09-19
- Bolt adds DeepSeek V4.1 Flash after week one: usage 10x DeepSeek V4 Pro — rohanpaul_ai · 2026-09-19
- Shared agent sessions: seeing what your collaborator is typing before it hits the agent — steipete · 2026-09-19
- npm co-founder Laurie Voss: agents are eating SDLC, product engineering pays $240k — round · 2026-09-19
- Tool-assisted AI clears Geometry Dash demon level Clubstep, methods open-sourced — imjustnewatai · 2026-09-19
- ty to ship full support for recursive implicit type aliases, built on mu types — charliermarsh · 2026-09-19