fix(channel): hot-apply runtime config updates for running channel service

docs/commands-reference.md

@@ -52,6 +52,7 @@ Last verified: **February 19, 2026**.
 - `zeroclaw service install`
 - `zeroclaw service start`
 - `zeroclaw service stop`
+- `zeroclaw service restart`
 - `zeroclaw service status`
 - `zeroclaw service uninstall`
 
@@ -90,6 +91,13 @@ Runtime in-chat commands (Telegram/Discord while channel server is running):
 - `/model`
 - `/model <model-id>`
 
+Channel runtime also watches `config.toml` and hot-applies updates to:
+- `default_provider`
+- `default_model`
+- `default_temperature`
+- `api_key` / `api_url` (for the default provider)
+- `reliability.*` provider retry settings
+
 `add/remove` currently route you back to managed setup/manual config paths (not full declarative mutators yet).
 
 ### `integrations`

docs/config-reference.md

@@ -349,6 +349,7 @@ Notes:
 - When a timeout occurs, users receive: `⚠️ Request timed out while waiting for the model. Please try again.`
 - Telegram-only interruption behavior is controlled with `channels_config.telegram.interrupt_on_new_message` (default `false`).
   When enabled, a newer message from the same sender in the same chat cancels the in-flight request and preserves interrupted user context.
+- While `zeroclaw channel start` is running, updates to `default_provider`, `default_model`, `default_temperature`, `api_key`, `api_url`, and `reliability.*` are hot-applied from `config.toml` on the next inbound message.
 
 See detailed channel matrix and allowlist behavior in [channels-reference.md](channels-reference.md).
 
@@ -425,6 +426,7 @@ After editing config:
 zeroclaw status
 zeroclaw doctor
 zeroclaw channel doctor
+zeroclaw service restart
 ```
 
 ## Related Docs