nagyist
diff --git a/‎CHANGELOG-next.md‎
Lines changed: 33 additions & 0 deletions b/‎CHANGELOG-next.md‎
Lines changed: 33 additions & 0 deletions
diff --git a/‎Cargo.lock‎
Lines changed: 2 additions & 0 deletions b/‎Cargo.lock‎
Lines changed: 2 additions & 0 deletions
diff --git a/‎benches/agent_benchmarks.rs‎
Lines changed: 2 additions & 2 deletions b/‎benches/agent_benchmarks.rs‎
Lines changed: 2 additions & 2 deletions
diff --git a/‎crates/zeroclaw-api/src/provider.rs‎
Lines changed: 118 additions & 21 deletions b/‎crates/zeroclaw-api/src/provider.rs‎
Lines changed: 118 additions & 21 deletions
@@ -113,6 +113,10 @@
 - Slack config: `channel_id` deprecated in favour of `channel_ids` (plural) for V2.
 - Nostr, WhatsApp Web, and hardware wizard sections wired into the onboarding flow
   (#5640).
+- Fresh `[providers.models.<name>]` entries no longer serialize family-irrelevant
+  default flags: `requires_openai_auth = false`, `merge_system_into_user = false`,
+  and empty `name`/`base_url`/`wire_api` entries are now omitted when they carry
+  their struct-level defaults. Existing configs deserialize unchanged.
 
 ### Observability
 
@@ -147,6 +151,27 @@
 - Fixed: `cron_run` tool output was not being delivered to configured channels.
 - Windows: the shell console window is now hidden when running as a background process
   (#5563).
+- `zeroclaw onboard` now defaults to the ratatui TUI backend. Pass `--cli` to force the
+  terminal-prompt backend; `--tui` is accepted as a deprecated no-op for one release.
+  Falls back to the terminal-prompt backend automatically when stdout isn't a TTY or
+  ratatui init fails.
+- `zeroclaw onboard` advanced-settings walk now filters out fields that don't apply to
+  the selected provider family — no more `azure_openai_*` prompts when configuring
+  Anthropic, no more `wire_api` / `requires_openai_auth` prompts outside the OpenAI
+  family.
+- `zeroclaw onboard` model-catalog fetch failures now name the provider in the fallback
+  note (`"Catalog lookup failed for <provider> — enter a model id manually"`) and log
+  the underlying error at `debug` level instead of swallowing it entirely.
+- `zeroclaw models` restored — routes to the same live model-probe output as
+  `zeroclaw doctor models`, preserving the `--provider` flag for `models list` /
+  `models refresh` subcommands.
+- Fixed: `TermUi::editor` close-without-save now returns `Answer::Back` (rewind)
+  instead of accepting the unchanged buffer silently. Matches the navigation contract
+  every other prompt method honors.
+- Fixed: `channels.mochat` now respects its `enabled` flag. Previously the orchestrator
+  registered the Mochat channel whenever the `[channels.mochat]` section existed,
+  producing `Mochat: poll request error: builder error` log spam on invalid or empty
+  URLs. Matches the WeCom / ClawdTalk enabled-gate pattern.
 
 ### Skills (Claude Code)
 
@@ -231,6 +256,14 @@ release but will not be supported indefinitely.
 Use `zeroclaw config` instead. The `props` subcommand still works and will not be
 removed in this release, but it will emit a deprecation notice.
 
+### `zeroclaw onboard` defaults to TUI
+
+`zeroclaw onboard` now launches the ratatui TUI backend by default. Users who were
+relying on the old terminal-prompt behavior should pass `--cli` explicitly. The
+previous `--tui` flag is accepted for one release as a deprecated no-op and emits a
+warning pointing at the new default. In non-TTY environments (piped output, CI without
+`--quick`) onboarding automatically falls back to the terminal-prompt backend.
+
 ### Slack `channel_id` deprecated
 
 Use `channel_ids` (a list) in the Slack config block. `channel_id` (singular) still
 
@@ -76,7 +76,7 @@ impl Provider for BenchProvider {
         _system_prompt: Option<&str>,
         _message: &str,
         _model: &str,
-        _temperature: f64,
+        _temperature: Option<f64>,
     ) -> Result<String> {
         Ok("fallback".into())
     }
@@ -85,7 +85,7 @@ impl Provider for BenchProvider {
         &self,
         _request: ChatRequest<'_>,
         _model: &str,
-        _temperature: f64,
+        _temperature: Option<f64>,
     ) -> Result<ChatResponse> {
         let mut guard = self.responses.lock().unwrap();
         if guard.is_empty() {
 
@@ -295,13 +295,70 @@ pub enum ToolsPayload {
     PromptGuided { instructions: String },
 }
 
+/// Industry-neutral sampling temperature. OpenAI, Gemini, OpenRouter, and
+/// most OpenAI-compatible endpoints document 0.7 as their typical default;
+/// Anthropic and Ollama override (1.0 and 0.0 respectively).
+pub const BASELINE_TEMPERATURE: f64 = 0.7;
+
+/// Output-token budget roomy enough for typical agent turns. Providers
+/// override per family where the model's own context window is the
+/// binding constraint.
+pub const BASELINE_MAX_TOKENS: u32 = 4096;
+
+/// HTTP timeout for cloud inference. Local providers (Ollama) override
+/// upward since CPU/GPU-bound inference runs slower than round-tripping to
+/// a hyperscaler.
+pub const BASELINE_TIMEOUT_SECS: u64 = 120;
+
+/// Wire protocol used when the provider doesn't declare one. Only OpenAI's
+/// Codex stack uses the "responses" protocol; everything else speaks the
+/// classic chat completions shape.
+pub const BASELINE_WIRE_API: &str = "chat_completions";
+
 #[async_trait]
 pub trait Provider: Send + Sync {
     /// Query provider capabilities.
     fn capabilities(&self) -> ProviderCapabilities {
         ProviderCapabilities::default()
     }
 
+    // ── Provider-family defaults ────────────────────────────────────────────
+    // Called by every chat/stream method's `temperature.unwrap_or(self.default_temperature())`
+    // and by `zeroclaw onboard` to prefill prompts with a visible default.
+    // Baselines are the industry-neutral fallback; override per family where
+    // the API docs disagree (Anthropic → 1.0, Ollama → 0.0 for deterministic
+    // local inference).
+
+    /// Temperature used when the caller passes `None`. Override per family.
+    fn default_temperature(&self) -> f64 {
+        BASELINE_TEMPERATURE
+    }
+
+    /// Max output tokens used when the caller / config doesn't set one.
+    fn default_max_tokens(&self) -> u32 {
+        BASELINE_MAX_TOKENS
+    }
+
+    /// HTTP timeout (seconds) used when the caller / config doesn't set one.
+    fn default_timeout_secs(&self) -> u64 {
+        BASELINE_TIMEOUT_SECS
+    }
+
+    /// Canonical public API endpoint, when there is one. Returned as a
+    /// string slice so provider impls can serve from `const &'static str`s
+    /// without allocations. `None` = provider has no universal endpoint
+    /// (local providers, auth-less CLIs, user-BYO endpoints).
+    fn default_base_url(&self) -> Option<&str> {
+        None
+    }
+
+    /// Wire protocol variant. Either `"responses"` (OpenAI Codex-style) or
+    /// `"chat_completions"` (everything else). Providers override to their
+    /// native format.
+    fn default_wire_api(&self) -> &str {
+        BASELINE_WIRE_API
+    }
+
     /// Convert tool specifications to provider-native format.
     fn convert_tools(&self, tools: &[ToolSpec]) -> ToolsPayload {
         ToolsPayload::PromptGuided {
@@ -310,31 +367,46 @@ pub trait Provider: Send + Sync {
     }
 
     /// Simple one-shot chat (single user message, no explicit system prompt).
+    ///
+    /// `temperature == None` means "use `self.default_temperature()`". The
+    /// unwrap lives inside each provider impl, not at every call site.
     async fn simple_chat(
         &self,
         message: &str,
         model: &str,
-        temperature: f64,
+        temperature: Option<f64>,
     ) -> anyhow::Result<String> {
         self.chat_with_system(None, message, model, temperature)
             .await
     }
 
-    /// One-shot chat with optional system prompt.
+    /// One-shot chat with optional system prompt. See `simple_chat` for
+    /// the `temperature` contract.
     async fn chat_with_system(
         &self,
         system_prompt: Option<&str>,
         message: &str,
         model: &str,
-        temperature: f64,
+        temperature: Option<f64>,
     ) -> anyhow::Result<String>;
 
-    /// Multi-turn conversation.
+    /// Fetch the list of available model IDs for this provider.
+    ///
+    /// Used by onboard to present a live model picker. Default bails with
+    /// "not supported"; concrete providers override to hit their own public
+    /// endpoint (OpenRouter, Ollama) or delegate to the shared models.dev
+    /// catalog (no auth required) in `zeroclaw_providers::models_dev`.
+    async fn list_models(&self) -> anyhow::Result<Vec<String>> {
+        anyhow::bail!("live model listing is not supported for this provider")
+    }
+
+    /// Multi-turn conversation. See `simple_chat` for the `temperature`
+    /// contract.
     async fn chat_with_history(
         &self,
         messages: &[ChatMessage],
         model: &str,
-        temperature: f64,
+        temperature: Option<f64>,
     ) -> anyhow::Result<String> {
         let system = messages
             .iter()
@@ -349,12 +421,13 @@ pub trait Provider: Send + Sync {
             .await
     }
 
-    /// Structured chat API for agent loop callers.
+    /// Structured chat API for agent loop callers. See `simple_chat` for
+    /// the `temperature` contract.
     async fn chat(
         &self,
         request: ChatRequest<'_>,
         model: &str,
-        temperature: f64,
+        temperature: Option<f64>,
     ) -> anyhow::Result<ChatResponse> {
         if let Some(tools) = request.tools
             && !tools.is_empty()
@@ -418,12 +491,13 @@ pub trait Provider: Send + Sync {
     }
 
     /// Chat with tool definitions for native function calling support.
+    /// See `simple_chat` for the `temperature` contract.
     async fn chat_with_tools(
         &self,
         messages: &[ChatMessage],
         _tools: &[serde_json::Value],
         model: &str,
-        temperature: f64,
+        temperature: Option<f64>,
     ) -> anyhow::Result<ChatResponse> {
         let text = self.chat_with_history(messages, model, temperature).await?;
         Ok(ChatResponse {
@@ -444,24 +518,26 @@ pub trait Provider: Send + Sync {
         false
     }
 
-    /// Streaming chat with optional system prompt.
+    /// Streaming chat with optional system prompt. See `simple_chat` for
+    /// the `temperature` contract.
     fn stream_chat_with_system(
         &self,
         _system_prompt: Option<&str>,
         _message: &str,
         _model: &str,
-        _temperature: f64,
+        _temperature: Option<f64>,
         _options: StreamOptions,
     ) -> stream::BoxStream<'static, StreamResult<StreamChunk>> {
         stream::empty().boxed()
     }
 
-    /// Streaming chat with history.
+    /// Streaming chat with history. See `simple_chat` for the `temperature`
+    /// contract.
     fn stream_chat_with_history(
         &self,
         messages: &[ChatMessage],
         model: &str,
-        temperature: f64,
+        temperature: Option<f64>,
         options: StreamOptions,
     ) -> stream::BoxStream<'static, StreamResult<StreamChunk>> {
         let system = messages
@@ -476,12 +552,13 @@ pub trait Provider: Send + Sync {
         self.stream_chat_with_system(system, last_user, model, temperature, options)
     }
 
-    /// Structured streaming chat interface.
+    /// Structured streaming chat interface. See `simple_chat` for the
+    /// `temperature` contract.
     fn stream_chat(
         &self,
         request: ChatRequest<'_>,
         model: &str,
-        temperature: f64,
+        temperature: Option<f64>,
         options: StreamOptions,
     ) -> stream::BoxStream<'static, StreamResult<StreamEvent>> {
         self.stream_chat_with_history(request.messages, model, temperature, options)
@@ -500,6 +577,26 @@ impl<T: Provider + ?Sized> Provider for Arc<T> {
         self.as_ref().capabilities()
     }
 
+    fn default_temperature(&self) -> f64 {
+        self.as_ref().default_temperature()
+    }
+
+    fn default_max_tokens(&self) -> u32 {
+        self.as_ref().default_max_tokens()
+    }
+
+    fn default_timeout_secs(&self) -> u64 {
+        self.as_ref().default_timeout_secs()
+    }
+
+    fn default_base_url(&self) -> Option<&str> {
+        self.as_ref().default_base_url()
+    }
+
+    fn default_wire_api(&self) -> &str {
+        self.as_ref().default_wire_api()
+    }
+
     fn convert_tools(&self, tools: &[ToolSpec]) -> ToolsPayload {
         self.as_ref().convert_tools(tools)
     }
@@ -517,7 +614,7 @@ impl<T: Provider + ?Sized> Provider for Arc<T> {
         system_prompt: Option<&str>,
         message: &str,
         model: &str,
-        temperature: f64,
+        temperature: Option<f64>,
     ) -> anyhow::Result<String> {
         self.as_ref()
             .chat_with_system(system_prompt, message, model, temperature)
@@ -528,7 +625,7 @@ impl<T: Provider + ?Sized> Provider for Arc<T> {
         &self,
         messages: &[ChatMessage],
         model: &str,
-        temperature: f64,
+        temperature: Option<f64>,
     ) -> anyhow::Result<String> {
         self.as_ref()
             .chat_with_history(messages, model, temperature)
@@ -539,7 +636,7 @@ impl<T: Provider + ?Sized> Provider for Arc<T> {
         &self,
         request: ChatRequest<'_>,
         model: &str,
-        temperature: f64,
+        temperature: Option<f64>,
     ) -> anyhow::Result<ChatResponse> {
         self.as_ref().chat(request, model, temperature).await
     }
@@ -553,7 +650,7 @@ impl<T: Provider + ?Sized> Provider for Arc<T> {
         messages: &[ChatMessage],
         tools: &[serde_json::Value],
         model: &str,
-        temperature: f64,
+        temperature: Option<f64>,
     ) -> anyhow::Result<ChatResponse> {
         self.as_ref()
             .chat_with_tools(messages, tools, model, temperature)
@@ -573,7 +670,7 @@ impl<T: Provider + ?Sized> Provider for Arc<T> {
         system_prompt: Option<&str>,
         message: &str,
         model: &str,
-        temperature: f64,
+        temperature: Option<f64>,
         options: StreamOptions,
     ) -> stream::BoxStream<'static, StreamResult<StreamChunk>> {
         self.as_ref()
@@ -584,7 +681,7 @@ impl<T: Provider + ?Sized> Provider for Arc<T> {
         &self,
         messages: &[ChatMessage],
         model: &str,
-        temperature: f64,
+        temperature: Option<f64>,
         options: StreamOptions,
     ) -> stream::BoxStream<'static, StreamResult<StreamChunk>> {
         self.as_ref()
@@ -595,7 +692,7 @@ impl<T: Provider + ?Sized> Provider for Arc<T> {
         &self,
         request: ChatRequest<'_>,
         model: &str,
-        temperature: f64,
+        temperature: Option<f64>,
         options: StreamOptions,
     ) -> stream::BoxStream<'static, StreamResult<StreamEvent>> {
         self.as_ref()