support configurable orchestrator model via orchestration config section

cli/planoai/config_generator.py

@@ -403,12 +403,26 @@ def validate_and_render_schema():
             }
         )
 
-    if "plano-orchestrator" not in model_provider_name_set:
+    orchestration_config = config_yaml.get("orchestration", {})
+    orchestration_model_provider = orchestration_config.get("llm_provider", None)
+
+    if (
+        orchestration_model_provider
+        and orchestration_model_provider not in model_provider_name_set
+    ):
+        raise Exception(
+            f"Orchestration llm_provider {orchestration_model_provider} is not defined in model_providers"
+        )
+
+    if (
+        orchestration_model_provider is None
+        and "plano-orchestrator" not in model_provider_name_set
+    ):
         updated_model_providers.append(
             {
                 "name": "plano-orchestrator",
                 "provider_interface": "arch",
-                "model": "Plano-Orchestrator",
+                "model": orchestration_config.get("model", "Plano-Orchestrator"),
                 "internal": True,
             }
         )

config/plano_config_schema.yaml

@@ -416,6 +416,14 @@ properties:
       model:
         type: string
     additionalProperties: false
+  orchestration:
+    type: object
+    properties:
+      llm_provider:
+        type: string
+      model:
+        type: string
+    additionalProperties: false
   state_storage:
     type: object
     properties:

crates/brightstaff/src/handlers/agent_selector.rs

@@ -178,6 +178,7 @@ mod tests {
         Arc::new(OrchestratorService::new(
             "http://localhost:8080".to_string(),
             "test-model".to_string(),
+            "plano-orchestrator".to_string(),
         ))
     }
 

crates/brightstaff/src/handlers/integration_tests.rs

@@ -23,6 +23,7 @@ mod tests {
         Arc::new(OrchestratorService::new(
             "http://localhost:8080".to_string(),
             "test-model".to_string(),
+            "plano-orchestrator".to_string(),
         ))
     }
 

crates/brightstaff/src/main.rs

@@ -11,9 +11,7 @@ use brightstaff::state::StateStorage;
 use brightstaff::utils::tracing::init_tracer;
 use bytes::Bytes;
 use common::configuration::{Agent, Configuration};
-use common::consts::{
-    CHAT_COMPLETIONS_PATH, MESSAGES_PATH, OPENAI_RESPONSES_API_PATH, PLANO_ORCHESTRATOR_MODEL_NAME,
-};
+use common::consts::{CHAT_COMPLETIONS_PATH, MESSAGES_PATH, OPENAI_RESPONSES_API_PATH};
 use common::llm_providers::LlmProviders;
 use http_body_util::{combinators::BoxBody, BodyExt, Empty};
 use hyper::body::Incoming;
@@ -35,6 +33,8 @@ pub mod router;
 const BIND_ADDRESS: &str = "0.0.0.0:9091";
 const DEFAULT_ROUTING_LLM_PROVIDER: &str = "arch-router";
 const DEFAULT_ROUTING_MODEL_NAME: &str = "Arch-Router";
+const DEFAULT_ORCHESTRATOR_LLM_PROVIDER: &str = "plano-orchestrator";
+const DEFAULT_ORCHESTRATOR_MODEL_NAME: &str = "Plano-Orchestrator";
 
 // Utility function to extract the context from the incoming request headers
 fn extract_context_from_request(req: &Request<Incoming>) -> Context {
@@ -109,9 +109,22 @@ async fn main() -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
         routing_llm_provider,
     ));
 
+    let orchestrator_model_name: String = plano_config
+        .orchestration
+        .as_ref()
+        .and_then(|o| o.model.clone())
+        .unwrap_or_else(|| DEFAULT_ORCHESTRATOR_MODEL_NAME.to_string());
+
+    let orchestrator_llm_provider: String = plano_config
+        .orchestration
+        .as_ref()
+        .and_then(|o| o.model_provider.clone())
+        .unwrap_or_else(|| DEFAULT_ORCHESTRATOR_LLM_PROVIDER.to_string());
+
     let orchestrator_service: Arc<OrchestratorService> = Arc::new(OrchestratorService::new(
         format!("{llm_provider_url}{CHAT_COMPLETIONS_PATH}"),
-        PLANO_ORCHESTRATOR_MODEL_NAME.to_string(),
+        orchestrator_model_name,
+        orchestrator_llm_provider,
     ));
 
     let model_aliases = Arc::new(plano_config.model_aliases.clone());

crates/brightstaff/src/router/plano_orchestrator.rs

@@ -2,7 +2,7 @@ use std::{collections::HashMap, sync::Arc};
 
 use common::{
     configuration::{AgentUsagePreference, OrchestrationPreference},
-    consts::{ARCH_PROVIDER_HINT_HEADER, PLANO_ORCHESTRATOR_MODEL_NAME, REQUEST_ID_HEADER},
+    consts::{ARCH_PROVIDER_HINT_HEADER, REQUEST_ID_HEADER},
 };
 use hermesllm::apis::openai::{ChatCompletionsResponse, Message};
 use hyper::header;
@@ -19,6 +19,7 @@ pub struct OrchestratorService {
     orchestrator_url: String,
     client: reqwest::Client,
     orchestrator_model: Arc<dyn OrchestratorModel>,
+    orchestrator_provider_name: String,
 }
 
 #[derive(Debug, Error)]
@@ -36,7 +37,11 @@ pub enum OrchestrationError {
 pub type Result<T> = std::result::Result<T, OrchestrationError>;
 
 impl OrchestratorService {
-    pub fn new(orchestrator_url: String, orchestration_model_name: String) -> Self {
+    pub fn new(
+        orchestrator_url: String,
+        orchestration_model_name: String,
+        orchestrator_provider_name: String,
+    ) -> Self {
         // Empty agent orchestrations - will be provided via usage_preferences in requests
         let agent_orchestrations: HashMap<String, Vec<OrchestrationPreference>> = HashMap::new();
 
@@ -50,6 +55,7 @@ impl OrchestratorService {
             orchestrator_url,
             client: reqwest::Client::new(),
             orchestrator_model,
+            orchestrator_provider_name,
         }
     }
 
@@ -91,7 +97,7 @@ impl OrchestratorService {
 
         orchestration_request_headers.insert(
             header::HeaderName::from_static(ARCH_PROVIDER_HINT_HEADER),
-            header::HeaderValue::from_str(PLANO_ORCHESTRATOR_MODEL_NAME).unwrap(),
+            header::HeaderValue::from_str(&self.orchestrator_provider_name).unwrap(),
         );
 
         // Inject OpenTelemetry trace context from current span
@@ -110,7 +116,7 @@ impl OrchestratorService {
 
         orchestration_request_headers.insert(
             header::HeaderName::from_static("model"),
-            header::HeaderValue::from_static(PLANO_ORCHESTRATOR_MODEL_NAME),
+            header::HeaderValue::from_str(&self.orchestrator_provider_name).unwrap(),
         );
 
         let start_time = std::time::Instant::now();

crates/common/src/configuration.rs

@@ -13,6 +13,12 @@ pub struct Routing {
     pub model: Option<String>,
 }
 
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct Orchestration {
+    pub model_provider: Option<String>,
+    pub model: Option<String>,
+}
+
 #[derive(Debug, Clone, Serialize, Deserialize)]
 pub struct ModelAlias {
     pub target: String,
@@ -73,6 +79,7 @@ pub struct Configuration {
     pub tracing: Option<Tracing>,
     pub mode: Option<GatewayMode>,
     pub routing: Option<Routing>,
+    pub orchestration: Option<Orchestration>,
     pub agents: Option<Vec<Agent>>,
     pub filters: Option<Vec<Agent>>,
     pub listeners: Vec<Listener>,

crates/common/src/consts.rs

@@ -33,5 +33,4 @@ pub const OTEL_COLLECTOR_HTTP: &str = "opentelemetry_collector_http";
 pub const LLM_ROUTE_HEADER: &str = "x-arch-llm-route";
 pub const ENVOY_RETRY_HEADER: &str = "x-envoy-max-retries";
 pub const BRIGHT_STAFF_SERVICE_NAME: &str = "brightstaff";
-pub const PLANO_ORCHESTRATOR_MODEL_NAME: &str = "Plano-Orchestrator";
 pub const ARCH_FC_CLUSTER: &str = "arch";