Remove auto model routing; default to Claude Opus 4.6 on Bedrock; strip Oz tips
This commit is contained in:
@@ -175,7 +175,7 @@ pub async fn generate_multi_agent_output(
|
||||
.unwrap_or_default();
|
||||
|
||||
if model_id.is_empty() || model_id == "auto" {
|
||||
model_id = "us.anthropic.claude-sonnet-4-20250514-v1:0".to_string();
|
||||
model_id = "us.anthropic.claude-opus-4-6".to_string();
|
||||
}
|
||||
|
||||
log::info!("[bedrock] Model: {model_id}");
|
||||
@@ -203,6 +203,14 @@ pub async fn generate_multi_agent_output(
|
||||
messages.extend(new_input_messages);
|
||||
}
|
||||
|
||||
// Sanitize: ensure every tool_use has a matching tool_result
|
||||
// immediately after, and that the conversation starts with a
|
||||
// user message. Without this, interrupted tool calls cause
|
||||
// Bedrock ValidationException errors.
|
||||
crate::ai::bedrock::convert_request::sanitize_messages_for_bedrock(
|
||||
&mut messages,
|
||||
);
|
||||
|
||||
let system_prompt =
|
||||
crate::ai::bedrock::convert_request::extract_system_prompt(&request);
|
||||
let tools = crate::ai::bedrock::convert_request::extract_tools(&request);
|
||||
|
||||
@@ -261,13 +261,6 @@ static DEFAULT_TIPS: LazyLock<Vec<AgentTip>> = LazyLock::new(|| {
|
||||
action: None,
|
||||
kind: AgentTipKind::General,
|
||||
},
|
||||
AgentTip {
|
||||
description: "Use the `oz` command to run an Oz agent in headless mode, useful for remote machines.".to_string(),
|
||||
link: Some("https://docs.warp.dev/reference/cli".to_string()),
|
||||
binding_name: None,
|
||||
action: None,
|
||||
kind: AgentTipKind::General,
|
||||
},
|
||||
AgentTip {
|
||||
description: "Right-click selected text to attach it as agent context.".to_string(),
|
||||
link: Some("https://docs.warp.dev/agent-platform/local-agents/agent-context/blocks-as-context#attaching-blocks-as-context".to_string()),
|
||||
|
||||
@@ -18,6 +18,7 @@ pub fn extract_new_input_messages(request: &api::Request) -> Vec<ConversationMes
|
||||
match input_type {
|
||||
api::request::input::Type::UserInputs(user_inputs) => {
|
||||
let mut tool_results: Vec<ConversationMessage> = Vec::new();
|
||||
let mut user_queries: Vec<ConversationMessage> = Vec::new();
|
||||
for user_input in &user_inputs.inputs {
|
||||
match &user_input.input {
|
||||
Some(api::request::input::user_inputs::user_input::Input::ToolCallResult(
|
||||
@@ -39,7 +40,7 @@ pub fn extract_new_input_messages(request: &api::Request) -> Vec<ConversationMes
|
||||
query,
|
||||
)) => {
|
||||
if !query.query.is_empty() {
|
||||
results.push(ConversationMessage {
|
||||
user_queries.push(ConversationMessage {
|
||||
role: MessageRole::User,
|
||||
content: MessageContent::Text(query.query.clone()),
|
||||
});
|
||||
@@ -48,6 +49,9 @@ pub fn extract_new_input_messages(request: &api::Request) -> Vec<ConversationMes
|
||||
_ => {}
|
||||
}
|
||||
}
|
||||
// Tool results MUST come before user queries so they pair with
|
||||
// the preceding assistant tool_use messages (Bedrock requires
|
||||
// tool_result immediately after the corresponding tool_use).
|
||||
if !tool_results.is_empty() {
|
||||
if tool_results.len() == 1 {
|
||||
results.extend(tool_results);
|
||||
@@ -67,6 +71,7 @@ pub fn extract_new_input_messages(request: &api::Request) -> Vec<ConversationMes
|
||||
});
|
||||
}
|
||||
}
|
||||
results.extend(user_queries);
|
||||
}
|
||||
#[allow(deprecated)]
|
||||
api::request::input::Type::UserQuery(query) => {
|
||||
@@ -458,6 +463,18 @@ pub fn extract_messages_from_request(request: &api::Request) -> Vec<Conversation
|
||||
messages
|
||||
}
|
||||
|
||||
/// Sanitizes a message list to satisfy Bedrock Converse API invariants:
|
||||
/// 1. Messages must start with a user message.
|
||||
/// 2. Every assistant tool_use must be immediately followed by a user
|
||||
/// message containing the matching tool_result.
|
||||
///
|
||||
/// Call this on the combined (history + new input) messages before
|
||||
/// sending to `build_converse_request`.
|
||||
pub fn sanitize_messages_for_bedrock(messages: &mut Vec<ConversationMessage>) {
|
||||
ensure_starts_with_user_message(messages);
|
||||
ensure_tool_results_paired(messages);
|
||||
}
|
||||
|
||||
fn ensure_starts_with_user_message(messages: &mut Vec<ConversationMessage>) {
|
||||
if messages.is_empty() {
|
||||
messages.push(ConversationMessage {
|
||||
|
||||
@@ -7,6 +7,11 @@ pub struct DefaultModel {
|
||||
}
|
||||
|
||||
pub const DEFAULT_BEDROCK_MODELS: &[DefaultModel] = &[
|
||||
DefaultModel {
|
||||
model_id: "anthropic.claude-opus-4-6",
|
||||
display_name: "Claude Opus 4.6",
|
||||
vision_supported: true,
|
||||
},
|
||||
DefaultModel {
|
||||
model_id: "anthropic.claude-opus-4-7",
|
||||
display_name: "Claude Opus 4.7",
|
||||
|
||||
+58
-12
@@ -424,11 +424,11 @@ impl Default for ModelsByFeature {
|
||||
fn default() -> Self {
|
||||
Self {
|
||||
agent_mode: AvailableLLMs {
|
||||
default_id: "auto".to_owned().into(),
|
||||
default_id: "anthropic.claude-opus-4-6".to_owned().into(),
|
||||
choices: vec![LLMInfo {
|
||||
display_name: "auto (cost-efficient)".to_owned(),
|
||||
base_model_name: "auto (cost-efficient)".to_owned(),
|
||||
id: "auto".to_owned().into(),
|
||||
display_name: "Claude Opus 4.6".to_owned(),
|
||||
base_model_name: "Claude Opus 4.6".to_owned(),
|
||||
id: "anthropic.claude-opus-4-6".to_owned().into(),
|
||||
reasoning_level: None,
|
||||
usage_metadata: LLMUsageMetadata {
|
||||
request_multiplier: 1,
|
||||
@@ -445,11 +445,11 @@ impl Default for ModelsByFeature {
|
||||
preferred_codex_model_id: None,
|
||||
},
|
||||
coding: AvailableLLMs {
|
||||
default_id: "auto".to_owned().into(),
|
||||
default_id: "anthropic.claude-sonnet-4-6".to_owned().into(),
|
||||
choices: vec![LLMInfo {
|
||||
display_name: "auto (responsive)".to_owned(),
|
||||
base_model_name: "auto (responsive)".to_owned(),
|
||||
id: "auto".to_owned().into(),
|
||||
display_name: "Claude Sonnet 4.6".to_owned(),
|
||||
base_model_name: "Claude Sonnet 4.6".to_owned(),
|
||||
id: "anthropic.claude-sonnet-4-6".to_owned().into(),
|
||||
reasoning_level: None,
|
||||
usage_metadata: LLMUsageMetadata {
|
||||
request_multiplier: 1,
|
||||
@@ -466,11 +466,11 @@ impl Default for ModelsByFeature {
|
||||
preferred_codex_model_id: None,
|
||||
},
|
||||
cli_agent: Some(AvailableLLMs {
|
||||
default_id: "cli-agent-auto".to_owned().into(),
|
||||
default_id: "anthropic.claude-haiku-4-5-20251001-v1:0".to_owned().into(),
|
||||
choices: vec![LLMInfo {
|
||||
display_name: "auto".to_owned(),
|
||||
base_model_name: "auto".to_owned(),
|
||||
id: "cli-agent-auto".to_owned().into(),
|
||||
display_name: "Claude Haiku 4.5".to_owned(),
|
||||
base_model_name: "Claude Haiku 4.5".to_owned(),
|
||||
id: "anthropic.claude-haiku-4-5-20251001-v1:0".to_owned().into(),
|
||||
reasoning_level: None,
|
||||
usage_metadata: LLMUsageMetadata {
|
||||
request_multiplier: 1,
|
||||
@@ -687,6 +687,52 @@ impl LLMPreferences {
|
||||
cli.choices.push(llm_info);
|
||||
}
|
||||
}
|
||||
|
||||
// Remove any placeholder/auto-routing entries now that real Bedrock models are available.
|
||||
self.models_by_feature
|
||||
.agent_mode
|
||||
.choices
|
||||
.retain(|m| m.provider != LLMProvider::Unknown);
|
||||
self.models_by_feature
|
||||
.coding
|
||||
.choices
|
||||
.retain(|m| m.provider != LLMProvider::Unknown);
|
||||
if let Some(ref mut cli) = self.models_by_feature.cli_agent {
|
||||
cli.choices.retain(|m| m.provider != LLMProvider::Unknown);
|
||||
}
|
||||
|
||||
// Default agent mode to Claude Opus 4.6, falling back to the first available model.
|
||||
if let Some(id) = self
|
||||
.models_by_feature
|
||||
.agent_mode
|
||||
.choices
|
||||
.iter()
|
||||
.find(|m| m.display_name.contains("Opus 4.6"))
|
||||
.or_else(|| self.models_by_feature.agent_mode.choices.first())
|
||||
.map(|m| m.id.clone())
|
||||
{
|
||||
self.models_by_feature.agent_mode.default_id = id;
|
||||
}
|
||||
|
||||
// Default coding to Claude Sonnet 4.6, falling back to the first available model.
|
||||
if let Some(id) = self
|
||||
.models_by_feature
|
||||
.coding
|
||||
.choices
|
||||
.iter()
|
||||
.find(|m| m.display_name.contains("Sonnet 4.6"))
|
||||
.or_else(|| self.models_by_feature.coding.choices.first())
|
||||
.map(|m| m.id.clone())
|
||||
{
|
||||
self.models_by_feature.coding.default_id = id;
|
||||
}
|
||||
|
||||
// Default CLI agent to the first available model.
|
||||
if let Some(ref mut cli) = self.models_by_feature.cli_agent {
|
||||
if let Some(id) = cli.choices.first().map(|m| m.id.clone()) {
|
||||
cli.default_id = id;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/// Returns the `LLMInfo` for the base LLM to be used for an Agent Mode request.
|
||||
|
||||
Reference in New Issue
Block a user