Debate: is Codex's harness replaceable with MCP plugins, and will local models win in 5 years?
pvncher · x · 2026-10-06
An X debate on the future of AI coding harnesses. Developer zeeg argued that Codex's harness provides little real functionality and that most coding harnesses will be superseded by specialized ones.
pvncher countered:
- Questioning what in the Codex harness can't be rebuilt as an MCP plugin — using the Responses API plus the compaction endpoint when context runs out
- Skeptical that everyone will use local models in 5 years; hosted serving likely offers better economics at a given intelligence level
- Core claim: build systems to benefit from the optimized building blocks the lab making the model provides
The exchange touches two industry fault lines: harness vs. plugin architecture, and local vs. hosted models.
Related event: Agent harness tier list sparks debate over compaction and vendor lock-in(15 posts)→
More from coding & agent
- dhh's AI-coded Campfire benchmark questioned: Rust got 398 tuning commits, Elixir got one pass — kristoph · 2026-10-06
- Ambient intelligence demo: Opus 5.5 agent shares your screen and knows what you're looking at — RileyRalmuto · 2026-10-06
- n8n Grew 10x Last Year Despite Being Declared Dead Three Times, CEO Says — aakashgupta · 2026-10-06
- Dev builds My AI Usage Canvas to turn local Copilot usage data into token trends and insights — DanWahlin · 2026-10-06
- Ethan Mollick: my AI workflow spans local Codex, cloud Dot and Claude tasks — explaining it is chaos — emollick · 2026-10-06
- Developer builds a tool that turns math equations into playable music — measure_plan · 2026-10-06