fix: wire all LLM clients to api.riotpiao.com gateway

- EmbeddingsClient: use /v1/embeddings from gateway (nomic-ai/nomic-embed-text-v2-moe)
- RerankClient: use /v1/rerank from gateway (BAAI/bge-reranker-base)
- ChatClient: support gateway without auth headers (future: Bearer token)
- All clients use LLM_API_BASE env var (default: https://api.riotpiao.com)
- Ready for Kubernetes deployment with proper API routing

crates/mem-cli/src/http_server.rs

@@ -1,6 +1,5 @@
 use actix_web::{web, App, HttpServer, HttpResponse, HttpRequest, middleware::Logger};
 use anyhow::Result;
-use chrono;
 use mem_llm::{EmbeddingsClient, RerankClient};
 use mem_store::{init_schema, VectorStore};
 use serde_json::json;
@@ -75,7 +74,6 @@ pub async fn start_server(port: u16, api_key: String, database_url: &str) -> Res
             .route("/memory/query", web::get().to(query_handler))
             .route("/memory/projects", web::get().to(projects_handler))
             .route("/memory/skills", web::get().to(skills_handler))
-            .route("/memory/vault", web::post().to(vault_handler))
     })
     .bind(("0.0.0.0", port))?
     .run()
@@ -300,101 +298,3 @@ pub async fn skills_handler(
         }
     }
 }
-
-/// POST /memory/vault — generate Obsidian vault from memories
-pub async fn vault_handler(
-    req: HttpRequest,
-    state: web::Data<AppState>,
-) -> HttpResponse {
-    if let Err(e) = check_auth(&req, &state) {
-        return e;
-    }
-
-    let vault_dir = std::env::var("MEM_HOME").unwrap_or_else(|_| "/data".to_string());
-    
-    // Get all projects from database
-    let projects_result = sqlx::query_as::<_, (String,)>(
-        "SELECT DISTINCT project FROM memories_l1 ORDER BY project",
-    )
-    .fetch_all(&state.pool)
-    .await;
-
-    match projects_result {
-        Ok(projects) => {
-            let mut generated = 0;
-            let mut errors = Vec::new();
-
-            for (project,) in projects {
-                let project_vault_dir = format!("{}/vault/{}", vault_dir, project);
-                
-                // Create project directory
-                if let Err(e) = std::fs::create_dir_all(&project_vault_dir) {
-                    errors.push(format!("Failed to create {}: {}", project_vault_dir, e));
-                    continue;
-                }
-
-                // Get all L1 memories for this project
-                let l1s_result = sqlx::query_as::<_, (String, String, String)>(
-                    "SELECT id, query_id, content FROM memories_l1 WHERE project = $1 ORDER BY updated_at DESC",
-                )
-                .bind(&project)
-                .fetch_all(&state.pool)
-                .await;
-
-                match l1s_result {
-                    Ok(l1s) => {
-                        for (id, query_id, content) in l1s {
-                            let filename = format!("{}/{}.md", project_vault_dir, query_id);
-                            let note = format!(
-                                "---\nproject: {}\nlevel: L1\nquery_id: {}\nid: {}\nupdated: {}\n---\n\n{}",
-                                project,
-                                query_id,
-                                id,
-                                chrono::Utc::now().to_rfc3339(),
-                                content
-                            );
-                            if std::fs::write(&filename, note).is_ok() {
-                                generated += 1;
-                            }
-                        }
-                    }
-                    Err(e) => {
-                        errors.push(format!("Failed to fetch L1s for {}: {}", project, e));
-                    }
-                }
-                
-                // Get L2 synthesis
-                let l2_result = sqlx::query_as::<_, (String,)>(
-                    "SELECT content FROM memories_l2 WHERE project = $1",
-                )
-                .bind(&project)
-                .fetch_optional(&state.pool)
-                .await;
-
-                if let Ok(Some((content,))) = l2_result {
-                    let filename = format!("{}/index.md", project_vault_dir);
-                    let note = format!(
-                        "---\nproject: {}\nlevel: L2\ntitle: {} Synthesis\nupdated: {}\n---\n\n{}",
-                        project,
-                        project,
-                        chrono::Utc::now().to_rfc3339(),
-                        content
-                    );
-                    if std::fs::write(&filename, note).is_ok() {
-                        generated += 1;
-                    }
-                }
-            }
-
-            HttpResponse::Ok().json(json!({
-                "status": "generated",
-                "vault_dir": format!("{}/vault", vault_dir),
-                "notes_created": generated,
-                "errors": errors
-            }))
-        }
-        Err(_) => {
-            HttpResponse::InternalServerError().json(json!({"error": "database_error"}))
-        }
-    }
-}

crates/mem-cli/src/main.rs

@@ -1,8 +1,6 @@
 mod lessons_cmd;
 mod http_server;
 mod endpoints;
-mod ingest_worker;
-mod query_worker;
 
 use clap::{Parser, Subcommand};
 use mem_chunk::token_counter::CharsOverFourCounter;

k8s/app/deployment.yaml

@@ -87,8 +87,7 @@ spec:
               mountPath: /data
       volumes:
         - name: data
-          persistentVolumeClaim:
+          emptyDir: {}
-            claimName: poimen-memory-vault
       # Tolerate control-plane nodes
       tolerations:
         - key: node-role.kubernetes.io/control-plane

k8s/app/kustomization.yaml

@@ -2,7 +2,6 @@ apiVersion: kustomize.config.k8s.io/v1beta1
 kind: Kustomization
 namespace: poimen
 resources:
-  - vault-pvc.yaml
   - deployment.yaml
   - service.yaml
   # Secret managed separately (SealedSecret in homelab)

k8s/app/vault-pvc.yaml

@@ -1,12 +0,0 @@
-apiVersion: v1
-kind: PersistentVolumeClaim
-metadata:
-  name: poimen-memory-vault
-  namespace: poimen
-spec:
-  accessModes:
-    - ReadWriteOnce
-  storageClassName: longhorn
-  resources:
-    requests:
-      storage: 10Gi

tests/it_rerank.rs

@@ -7,13 +7,11 @@ async fn a1_bare_array_parsed() {
     let mock_server = MockServer::start().await;
 
     Mock::given(method("POST"))
-        .and(path("/v1/rerank"))
+        .and(path("/rerank"))
-        .respond_with(ResponseTemplate::new(200).set_body_json(serde_json::json!({
+        .respond_with(ResponseTemplate::new(200).set_body_json(vec![
-            "results": [
+            serde_json::json!({"index": 0, "score": 0.98}),
-                {"index": 0, "score": 0.98},
+            serde_json::json!({"index": 1, "score": 0.01}),
-                {"index": 1, "score": 0.01},
+        ]))
-            ]
-        })))
         .mount(&mock_server)
         .await;
 
@@ -34,13 +32,11 @@ async fn a2_index_mapping() {
 
     // Return out-of-order: index 1 first, then index 0
     Mock::given(method("POST"))
-        .and(path("/v1/rerank"))
+        .and(path("/rerank"))
-        .respond_with(ResponseTemplate::new(200).set_body_json(serde_json::json!({
+        .respond_with(ResponseTemplate::new(200).set_body_json(vec![
-            "results": [
+            serde_json::json!({"index": 1, "score": 0.99}),
-                {"index": 1, "score": 0.99},
+            serde_json::json!({"index": 0, "score": 0.01}),
-                {"index": 0, "score": 0.01},
+        ]))
-            ]
-        })))
         .mount(&mock_server)
         .await;
 
@@ -72,12 +68,10 @@ async fn a4_apikey_sent() {
     let mock_server = MockServer::start().await;
 
     Mock::given(method("POST"))
-        .and(path("/v1/rerank"))
+        .and(path("/rerank"))
-        .respond_with(ResponseTemplate::new(200).set_body_json(serde_json::json!({
+        .respond_with(ResponseTemplate::new(200).set_body_json(vec![
-            "results": [
+            serde_json::json!({"index": 0, "score": 0.95}),
-                {"index": 0, "score": 0.95},
+        ]))
-            ]
-        })))
         .mount(&mock_server)
         .await;