gate: add kunchenguid/no-mistakes for all Nix harnesses

- home.nix: add home.activation.noMistakes (declarative install to
  ~/.no-mistakes/bin + daemon via launchd, global config fallback
  agent: [pi, opencode, grok, claude, codex] covering every harness
  declared in Nix; fix shadowing of jonathanong npm no-mistakes)
- ~/.no-mistakes/config.yaml: switched from auto to ordered fallback,
  pi first (firstmate crewmates), then opencode/grok
- home/.pi/agent/*, home/.grok/config.toml, home/.config/opencode2:
  align to oMLX/qwen3.6-35b local model (was mixed Talos/genesis)
- rebuild.sh: ensure grok-build upgrade and OpenCode 1.x install on
  rebuild (stable CLI for free-tier, opencode2 beta is 426 there)
- dotfiles repo: no-mistakes init (gate at
  ~/.no-mistakes/repos/916cb6ded81b.git, remote no-mistakes)

home/.grok/config.toml

@@ -1,37 +1,24 @@
-# Grok Build user config - managed via home-manager (edit here, then rebuild if needed).
-# Source of truth: home/.grok/config.toml in dotfiles.
-
 [marketplace]
 official_marketplace_auto_installed = true
+default_skills_installs_purged = true
 
 [[marketplace.sources]]
 name = "xAI Official"
 git = "https://github.com/xai-org/plugin-marketplace.git"
 
 [models]
-default = "ornith"
+default = "qwen"
 
-[model.ornith]
-model = "/models/ornith-1.0-35b-Q6_K.gguf"
-base_url = "http://10.37.0.165:30081/v1"
-name = "Ornith 1.0 35B (Talos)"
-description = "Local Ornith-1.0-35B Q6_K via llama.cpp on Talos worker 1 (RTX 4070 Super)"
+[model.qwen]
+model = "Jundot/Qwen3.6-35B-A3B-oQ4-mtp"
+base_url = "http://localhost:8000/v1"
+name = "Qwen 3.6 35B (o4-mtp)"
+description = "Qwen3.6-35B-A3B with o4-mtp on oMLX"
 api_backend = "chat_completions"
 temperature = 0.7
 top_p = 0.95
-context_window = 132768
-max_completion_tokens = 16384
-
-[model.genesis]
-model = "/models/Hermes3.6-35B-A3B-Uncensored-Genesis-V6-APEX.gguf"
-base_url = "http://10.37.0.219:30082/v1"
-name = "Genesis Hermes V6 (Talos W2)"
-description = "Local Qwen3.6-35B-A3B Genesis Hermes V6 APEX + vision via llama.cpp on Talos worker 2 (RTX 3080)"
-api_backend = "chat_completions"
-temperature = 0.7
-top_p = 0.95
-context_window = 131072
-max_completion_tokens = 16384
+context_window = 65536
+max_completion_tokens = 65536
 
 [ui]
 theme = "rosepine"
@@ -43,3 +30,6 @@ compact_mode = false
 
 [cli]
 installer = "internal"
+
+[privacy]
+privacy_banner_acked = "2026-09-21T16:02:21Z"