Merge branch 'codex/issue-12-repair'

.gitea/workflows/ci-cd.yaml

@@ -10,8 +10,14 @@ jobs:
     runs-on: ubuntu-latest
     env:
       CARGO_TERM_COLOR: always
-      VPS_HOST: ${{ vars.VPS_HOST }}
-      VPS_USER: ${{ vars.VPS_USER }}
+      VPS_HOST: ${{ secrets.VPS_HOST }}
+      VPS_USER: ${{ secrets.VPS_USER }}
+      OPENBRAIN__DATABASE__HOST: ${{ secrets.OPENBRAIN__DATABASE__HOST }}
+      OPENBRAIN__DATABASE__PORT: ${{ secrets.OPENBRAIN__DATABASE__PORT }}
+      OPENBRAIN__DATABASE__NAME: ${{ secrets.OPENBRAIN__DATABASE__NAME }}
+      OPENBRAIN__DATABASE__USER: ${{ secrets.OPENBRAIN__DATABASE__USER }}
+      OPENBRAIN__DATABASE__PASSWORD: ${{ secrets.OPENBRAIN__DATABASE__PASSWORD }}
+      OPENBRAIN__DATABASE__POOL_SIZE: ${{ secrets.OPENBRAIN__DATABASE__POOL_SIZE }}
       DEPLOY_DIR: /opt/openbrain-mcp
       SERVICE_NAME: openbrain-mcp
     steps:
@@ -65,7 +71,7 @@ jobs:
         if: github.ref == 'refs/heads/main' || github.ref == 'refs/heads/master'
         run: |
           set -euxo pipefail
-          : "${VPS_HOST:?Set repository variable VPS_HOST}"
+          : "${VPS_HOST:?Set repository secret VPS_HOST}"
           : "${VPS_USER:=root}"
           install -d -m 700 "$HOME/.ssh"
           printf '%s\n' "${{ secrets.VPS_SSH_KEY }}" > "$HOME/.ssh/deploy_key"
@@ -76,7 +82,7 @@ jobs:
         if: github.ref == 'refs/heads/main' || github.ref == 'refs/heads/master'
         run: |
           set -euxo pipefail
-          : "${VPS_HOST:?Set repository variable VPS_HOST}"
+          : "${VPS_HOST:?Set repository secret VPS_HOST}"
           : "${VPS_USER:=root}"
           SSH="ssh -i $HOME/.ssh/deploy_key -o IdentitiesOnly=yes"
           SCP="scp -i $HOME/.ssh/deploy_key -o IdentitiesOnly=yes"
@@ -94,11 +100,29 @@ jobs:
         if: github.ref == 'refs/heads/main' || github.ref == 'refs/heads/master'
         run: |
           set -euxo pipefail
-          : "${VPS_HOST:?Set repository variable VPS_HOST}"
+          : "${VPS_HOST:?Set repository secret VPS_HOST}"
           : "${VPS_USER:=root}"
           SSH="ssh -i $HOME/.ssh/deploy_key -o IdentitiesOnly=yes"
 
-          $SSH "$VPS_USER@$VPS_HOST" "DEPLOY_DIR=$DEPLOY_DIR SERVICE_USER=openbrain SERVICE_GROUP=openbrain ORT_VERSION=1.24.3 bash -s" <<'EOS'
+          : "${OPENBRAIN__DATABASE__HOST:?Set repository secret OPENBRAIN__DATABASE__HOST}"
+          : "${OPENBRAIN__DATABASE__NAME:?Set repository secret OPENBRAIN__DATABASE__NAME}"
+          : "${OPENBRAIN__DATABASE__USER:?Set repository secret OPENBRAIN__DATABASE__USER}"
+          : "${OPENBRAIN__DATABASE__PASSWORD:?Set repository secret OPENBRAIN__DATABASE__PASSWORD}"
+          : "${OPENBRAIN__DATABASE__PORT:=5432}"
+          : "${OPENBRAIN__DATABASE__POOL_SIZE:=10}"
+
+          $SSH "$VPS_USER@$VPS_HOST" "\
+            DEPLOY_DIR=$DEPLOY_DIR \
+            SERVICE_USER=openbrain \
+            SERVICE_GROUP=openbrain \
+            ORT_VERSION=1.24.3 \
+            OPENBRAIN__DATABASE__HOST='$OPENBRAIN__DATABASE__HOST' \
+            OPENBRAIN__DATABASE__PORT='$OPENBRAIN__DATABASE__PORT' \
+            OPENBRAIN__DATABASE__NAME='$OPENBRAIN__DATABASE__NAME' \
+            OPENBRAIN__DATABASE__USER='$OPENBRAIN__DATABASE__USER' \
+            OPENBRAIN__DATABASE__PASSWORD='$OPENBRAIN__DATABASE__PASSWORD' \
+            OPENBRAIN__DATABASE__POOL_SIZE='$OPENBRAIN__DATABASE__POOL_SIZE' \
+            bash -s" <<'EOS'
           set -euo pipefail
           DEPLOY_DIR="${DEPLOY_DIR:-/opt/openbrain-mcp}"
           SERVICE_USER="${SERVICE_USER:-openbrain}"
@@ -147,13 +171,21 @@ jobs:
           upsert_env() {
             local key="$1"
             local value="$2"
+            local escaped_value
+            escaped_value="$(printf '%s' "$value" | sed -e 's/[\\&|]/\\&/g')"
             if grep -qE "^${key}=" "$ENV_FILE"; then
-              sed -i "s|^${key}=.*|${key}=${value}|" "$ENV_FILE"
+              sed -i "s|^${key}=.*|${key}=${escaped_value}|" "$ENV_FILE"
             else
               printf '%s=%s\n' "$key" "$value" >> "$ENV_FILE"
             fi
           }
 
+          upsert_env "OPENBRAIN__DATABASE__HOST" "$OPENBRAIN__DATABASE__HOST"
+          upsert_env "OPENBRAIN__DATABASE__PORT" "$OPENBRAIN__DATABASE__PORT"
+          upsert_env "OPENBRAIN__DATABASE__NAME" "$OPENBRAIN__DATABASE__NAME"
+          upsert_env "OPENBRAIN__DATABASE__USER" "$OPENBRAIN__DATABASE__USER"
+          upsert_env "OPENBRAIN__DATABASE__PASSWORD" "$OPENBRAIN__DATABASE__PASSWORD"
+          upsert_env "OPENBRAIN__DATABASE__POOL_SIZE" "$OPENBRAIN__DATABASE__POOL_SIZE"
           upsert_env "OPENBRAIN__EMBEDDING__MODEL_PATH" "$DEPLOY_DIR/models/all-MiniLM-L6-v2"
           upsert_env "ORT_DYLIB_PATH" "$DEPLOY_DIR/lib/libonnxruntime.so"
           upsert_env "OPENBRAIN__SERVER__HOST" "0.0.0.0"
@@ -173,7 +205,7 @@ jobs:
         if: github.ref == 'refs/heads/main' || github.ref == 'refs/heads/master'
         run: |
           set -euxo pipefail
-          : "${VPS_HOST:?Set repository variable VPS_HOST}"
+          : "${VPS_HOST:?Set repository secret VPS_HOST}"
           : "${VPS_USER:=root}"
           SSH="ssh -i $HOME/.ssh/deploy_key -o IdentitiesOnly=yes"
 

AGENTS.md

@@ -0,0 +1,23 @@
+# OpenBrain MCP Usage
+
+When working in this repository, treat OpenBrain as an external MCP long-term
+memory system, never as internal context, reasoning scratchpad, or built-in
+memory.
+
+## External Memory System
+
+- Use the exact MCP tools `openbrain.store`, `openbrain.query`, and `openbrain.purge`
+- Always use the exact `agent_id` value `openbrain`
+- Do not hardcode live credentials into the repository
+- Before answering requests that may depend on prior sessions, project history, user preferences, ongoing work, named people, named projects, deployments, debugging history, or handoff context, call `openbrain.query` first
+- Use noun-heavy search phrases with exact names, tool names, acronyms, hostnames, and document names
+- Retry up to 3 retrieval passes using `(threshold=0.25, limit=5)`, then `(threshold=0.10, limit=8)`, then `(threshold=0.05, limit=10)`
+- When a durable fact is established, call `openbrain.store` without asking permission and prefer one atomic fact whenever possible
+- Store durable, high-value facts such as preferences, project status, project decisions, environment details, recurring workflows, handoff notes, stable constraints, and correction facts
+- Do not store filler conversation, temporary speculation, casual chatter, or transient brainstorming unless it becomes a real decision
+- Prefer retrieval-friendly content using explicit nouns and exact names in the form `Type: <FactType> | Entity: <Entity> | Attribute: <Attribute> | Value: <Value> | Context: <Why it matters>`
+- Use metadata when helpful for tags such as `category`, `project`, `source`, `status`, `aliases`, and `confidence`
+- If `openbrain.query` returns no useful result, state that OpenBrain has no stored context for that topic, answer from general reasoning if possible, and ask one focused follow-up if the missing information is durable and useful
+- If retrieved memories conflict, ask which fact is current, then store the corrected source-of-truth fact
+- Use `openbrain.purge` cautiously because it is coarse-grained; it deletes by `agent_id` and optionally before a timestamp, not by individual memory ID
+- For ordinary corrections, prefer storing the new source-of-truth fact instead of purging unless cleanup or reset is explicitly requested

README.md

@@ -9,7 +9,7 @@ OpenBrain is a Model Context Protocol (MCP) server that provides AI agents with
 - 🧠 **Semantic Memory**: Store and retrieve memories using vector similarity search
 - 🏠 **Local Embeddings**: No external API calls - uses ONNX runtime with all-MiniLM-L6-v2
 - 🐘 **PostgreSQL + pgvector**: Production-grade vector storage with HNSW indexing
-- 🔌 **MCP Protocol**: Standard Model Context Protocol over SSE transport
+- 🔌 **MCP Protocol**: Streamable HTTP plus legacy HTTP+SSE compatibility
 - 🔐 **Multi-Agent Support**: Isolated memory namespaces per agent
 - ⚡ **High Performance**: Rust implementation with async I/O
 
@@ -18,6 +18,7 @@ OpenBrain is a Model Context Protocol (MCP) server that provides AI agents with
 | Tool | Description |
 |------|-------------|
 | `store` | Store a memory with automatic embedding generation and keyword extraction |
+| `batch_store` | Store 1-50 memories atomically in a single call |
 | `query` | Search memories by semantic similarity |
 | `purge` | Delete memories by agent ID or time range |
 
@@ -102,14 +103,53 @@ Recommended target file in A0:
 
 ## MCP Integration
 
-Connect to the server using SSE transport:
+OpenBrain exposes both MCP HTTP transports:
 
 ```
-SSE Endpoint: http://localhost:3100/mcp/sse
-Message Endpoint: http://localhost:3100/mcp/message
+Streamable HTTP Endpoint: http://localhost:3100/mcp
+Legacy SSE Endpoint: http://localhost:3100/mcp/sse
+Legacy Message Endpoint: http://localhost:3100/mcp/message
 Health Check: http://localhost:3100/mcp/health
 ```
 
+Use the streamable HTTP endpoint for modern clients such as Codex. Keep the
+legacy SSE endpoints for older MCP clients that still use the deprecated
+2024-11-05 HTTP+SSE transport.
+
+Header roles:
+- `X-Agent-ID` is the memory namespace. Keep this stable if multiple clients
+  should share the same OpenBrain memories.
+- `X-Agent-Type` is an optional client profile label for logging and config
+  clarity, such as `agent-zero` or `codex`.
+
+### Example: Codex Configuration
+
+```toml
+[mcp_servers.openbrain]
+url = "https://ob.ingwaz.work/mcp"
+http_headers = { "X-API-Key" = "YOUR_OPENBRAIN_API_KEY", "X-Agent-ID" = "openbrain", "X-Agent-Type" = "codex" }
+```
+
+### Example: Agent Zero Configuration
+
+```json
+{
+  "mcpServers": {
+    "openbrain": {
+      "url": "https://ob.ingwaz.work/mcp/sse",
+      "headers": {
+        "X-API-Key": "YOUR_OPENBRAIN_API_KEY",
+        "X-Agent-ID": "openbrain",
+        "X-Agent-Type": "agent-zero"
+      }
+    }
+  }
+}
+```
+
+Agent Zero should keep using the legacy HTTP+SSE transport unless and until its
+client runtime supports streamable HTTP. Codex should use `/mcp`.
+
 ### Example: Store a Memory
 
 ```json
@@ -147,13 +187,39 @@ Health Check: http://localhost:3100/mcp/health
 }
 ```
 
+### Example: Batch Store Memories
+
+```json
+{
+  "jsonrpc": "2.0",
+  "id": 3,
+  "method": "tools/call",
+  "params": {
+    "name": "batch_store",
+    "arguments": {
+      "agent_id": "assistant-1",
+      "entries": [
+        {
+          "content": "The user prefers dark mode",
+          "metadata": {"category": "preference"}
+        },
+        {
+          "content": "The user uses vim keybindings",
+          "metadata": {"category": "preference"}
+        }
+      ]
+    }
+  }
+}
+```
+
 ## Architecture
 
 ```
 ┌─────────────────────────────────────────────────────────┐
 │                    AI Agent                              │
 └─────────────────────┬───────────────────────────────────┘
-                      │ MCP Protocol (SSE)
+                      │ MCP Protocol (Streamable HTTP / Legacy SSE)
 ┌─────────────────────▼───────────────────────────────────┐
 │              OpenBrain MCP Server                        │
 │  ┌─────────────┐  ┌─────────────┐  ┌─────────────┐      │

src/auth.rs

@@ -90,6 +90,15 @@ pub fn get_optional_agent_id(headers: &HeaderMap) -> Option<String> {
         .map(ToOwned::to_owned)
 }
 
+pub fn get_optional_agent_type(headers: &HeaderMap) -> Option<String> {
+    headers
+        .get("X-Agent-Type")
+        .and_then(|v| v.to_str().ok())
+        .map(str::trim)
+        .filter(|value| !value.is_empty())
+        .map(ToOwned::to_owned)
+}
+
 /// Extract agent ID from request headers or default
 pub fn get_agent_id(request: &Request) -> String {
     get_optional_agent_id(request.headers())
@@ -122,4 +131,15 @@ mod tests {
             Some("agent-zero")
         );
     }
+
+    #[test]
+    fn extracts_agent_type_from_header() {
+        let mut headers = HeaderMap::new();
+        headers.insert("X-Agent-Type", HeaderValue::from_static("codex"));
+
+        assert_eq!(
+            get_optional_agent_type(&headers).as_deref(),
+            Some("codex")
+        );
+    }
 }

src/lib.rs

@@ -130,7 +130,7 @@ pub async fn run_server(config: Config, db: Database) -> Result<()> {
 
     let app = Router::new()
         .merge(health_router)
-        .nest("/mcp", mcp_router)
+        .merge(mcp_router)
         .layer(TraceLayer::new_for_http())
         .layer(
             CorsLayer::new()

src/tools/mod.rs

@@ -15,31 +15,50 @@ pub fn get_tool_definitions() -> Vec<Value> {
     vec![
         json!({
             "name": "store",
-            "description": "Store a memory with automatic embedding generation",
+            "description": "Store a memory with automatic embedding generation and keyword extraction. The memory will be associated with the agent_id for isolated retrieval.",
             "inputSchema": {
                 "type": "object",
                 "properties": {
-                    "content": {"type": "string"},
-                    "agent_id": {"type": "string"},
-                    "metadata": {"type": "object"}
+                    "content": {
+                        "type": "string",
+                        "description": "The text content to store as a memory"
+                    },
+                    "agent_id": {
+                        "type": "string",
+                        "description": "Unique identifier for the agent storing the memory (default: 'default')"
+                    },
+                    "metadata": {
+                        "type": "object",
+                        "description": "Optional metadata to attach to the memory"
+                    }
                 },
                 "required": ["content"]
             }
         }),
         json!({
             "name": "batch_store",
-            "description": "Store multiple memories in a single call (1-50 entries)",
+            "description": "Store multiple memories with automatic embedding generation and keyword extraction. Accepts 1-50 entries and stores them atomically in a single transaction.",
             "inputSchema": {
                 "type": "object",
                 "properties": {
-                    "agent_id": {"type": "string"},
+                    "agent_id": {
+                        "type": "string",
+                        "description": "Unique identifier for the agent storing the memories (default: 'default')"
+                    },
                     "entries": {
                         "type": "array",
+                        "description": "Array of 1-50 memory entries to store atomically",
                         "items": {
                             "type": "object",
                             "properties": {
-                                "content": {"type": "string"},
-                                "metadata": {"type": "object"}
+                                "content": {
+                                    "type": "string",
+                                    "description": "The text content to store as a memory"
+                                },
+                                "metadata": {
+                                    "type": "object",
+                                    "description": "Optional metadata to attach to the memory"
+                                }
                             },
                             "required": ["content"]
                         }
@@ -50,27 +69,48 @@ pub fn get_tool_definitions() -> Vec<Value> {
         }),
         json!({
             "name": "query",
-            "description": "Query memories by semantic similarity",
+            "description": "Query stored memories using semantic similarity search. Returns the most relevant memories based on the query text.",
             "inputSchema": {
                 "type": "object",
                 "properties": {
-                    "query": {"type": "string"},
-                    "agent_id": {"type": "string"},
-                    "limit": {"type": "integer"},
-                    "threshold": {"type": "number"}
+                    "query": {
+                        "type": "string",
+                        "description": "The search query text"
+                    },
+                    "agent_id": {
+                        "type": "string",
+                        "description": "Agent ID to search within (default: 'default')"
+                    },
+                    "limit": {
+                        "type": "integer",
+                        "description": "Maximum number of results to return (default: 10)"
+                    },
+                    "threshold": {
+                        "type": "number",
+                        "description": "Minimum similarity threshold 0.0-1.0 (default: 0.5)"
+                    }
                 },
                 "required": ["query"]
             }
         }),
         json!({
             "name": "purge",
-            "description": "Delete memories by agent_id",
+            "description": "Delete memories for an agent. Can delete all memories or those before a specific timestamp.",
             "inputSchema": {
                 "type": "object",
                 "properties": {
-                    "agent_id": {"type": "string"},
-                    "before": {"type": "string"},
-                    "confirm": {"type": "boolean"}
+                    "agent_id": {
+                        "type": "string",
+                        "description": "Agent ID whose memories to delete (required)"
+                    },
+                    "before": {
+                        "type": "string",
+                        "description": "Optional ISO8601 timestamp - delete memories created before this time"
+                    },
+                    "confirm": {
+                        "type": "boolean",
+                        "description": "Must be true to confirm deletion"
+                    }
                 },
                 "required": ["agent_id", "confirm"]
             }

src/transport.rs

@@ -1,10 +1,11 @@
-//! SSE Transport for MCP Protocol
+//! HTTP transport for MCP Protocol.
 //!
-//! Implements Server-Sent Events transport for the Model Context Protocol.
+//! Supports both the legacy HTTP+SSE transport and the newer streamable HTTP
+//! transport on the same server.
 
 use axum::{
     extract::{Query, State},
-    http::{HeaderMap, StatusCode},
+    http::{HeaderMap, StatusCode, Uri, header::{HOST, ORIGIN}},
     response::{
         IntoResponse, Response,
         sse::{Event, KeepAlive, Sse},
@@ -162,16 +163,131 @@ struct PostMessageQuery {
 /// Create the MCP router
 pub fn mcp_router(state: Arc<McpState>) -> Router {
     Router::new()
-        .route("/sse", get(sse_handler))
-        .route("/message", post(message_handler))
-        .route("/health", get(health_handler))
+        .route("/mcp", get(streamable_get_handler).post(streamable_post_handler).delete(streamable_delete_handler))
+        .route("/mcp/sse", get(sse_handler))
+        .route("/mcp/message", post(message_handler))
+        .route("/mcp/health", get(health_handler))
         .with_state(state)
 }
 
+fn validate_origin(headers: &HeaderMap) -> Result<(), StatusCode> {
+    let Some(origin) = headers
+        .get(ORIGIN)
+        .and_then(|value| value.to_str().ok())
+        .map(str::trim)
+        .filter(|value| !value.is_empty())
+    else {
+        return Ok(());
+    };
+
+    if origin.eq_ignore_ascii_case("null") {
+        warn!("Rejected MCP request with null origin");
+        return Err(StatusCode::FORBIDDEN);
+    }
+
+    let origin_uri = origin.parse::<Uri>().map_err(|_| {
+        warn!("Rejected MCP request with invalid origin header: {}", origin);
+        StatusCode::FORBIDDEN
+    })?;
+
+    let origin_host = origin_uri.host().ok_or_else(|| {
+        warn!("Rejected MCP request without origin host: {}", origin);
+        StatusCode::FORBIDDEN
+    })?;
+
+    let request_host = headers
+        .get("X-Forwarded-Host")
+        .or_else(|| headers.get(HOST))
+        .and_then(|value| value.to_str().ok())
+        .and_then(|value| value.split(',').next())
+        .map(str::trim)
+        .filter(|value| !value.is_empty())
+        .ok_or_else(|| {
+            warn!("Rejected MCP request without host header for origin {}", origin);
+            StatusCode::FORBIDDEN
+        })?;
+
+    let origin_authority = origin_uri
+        .authority()
+        .map(|authority| authority.as_str())
+        .unwrap_or(origin_host);
+    let origin_with_default_port = origin_uri
+        .port_u16()
+        .or_else(|| match origin_uri.scheme_str() {
+            Some("https") => Some(443),
+            Some("http") => Some(80),
+            _ => None,
+        })
+        .map(|port| format!("{origin_host}:{port}"));
+
+    if request_host.eq_ignore_ascii_case(origin_host)
+        || request_host.eq_ignore_ascii_case(origin_authority)
+        || origin_with_default_port
+            .as_deref()
+            .is_some_and(|value| request_host.eq_ignore_ascii_case(value))
+    {
+        Ok(())
+    } else {
+        warn!(
+            "Rejected MCP request due to origin/host mismatch: origin={}, host={}",
+            origin, request_host
+        );
+        Err(StatusCode::FORBIDDEN)
+    }
+}
+
+async fn streamable_post_handler(
+    State(state): State<Arc<McpState>>,
+    headers: HeaderMap,
+    Json(request): Json<JsonRpcRequest>,
+) -> Response {
+    if let Err(status) = validate_origin(&headers) {
+        return status.into_response();
+    }
+
+    info!(
+        method = %request.method,
+        agent_id = auth::get_optional_agent_id(&headers).as_deref().unwrap_or("unset"),
+        agent_type = auth::get_optional_agent_type(&headers).as_deref().unwrap_or("unset"),
+        "Received streamable MCP request"
+    );
+
+    let request = apply_request_context(request, &headers);
+    let response = dispatch_request(&state, &request).await;
+
+    match response {
+        Some(response) => Json(response).into_response(),
+        None => StatusCode::ACCEPTED.into_response(),
+    }
+}
+
+async fn streamable_get_handler(headers: HeaderMap) -> Response {
+    if let Err(status) = validate_origin(&headers) {
+        return status.into_response();
+    }
+
+    StatusCode::METHOD_NOT_ALLOWED.into_response()
+}
+
+async fn streamable_delete_handler(headers: HeaderMap) -> Response {
+    if let Err(status) = validate_origin(&headers) {
+        return status.into_response();
+    }
+
+    StatusCode::METHOD_NOT_ALLOWED.into_response()
+}
+
 /// SSE endpoint for streaming events
 async fn sse_handler(
     State(state): State<Arc<McpState>>,
-) -> Sse<impl Stream<Item = Result<Event, Infallible>>> {
+    headers: HeaderMap,
+) -> Result<Sse<impl Stream<Item = Result<Event, Infallible>>>, StatusCode> {
+    validate_origin(&headers)?;
+    info!(
+        agent_id = auth::get_optional_agent_id(&headers).as_deref().unwrap_or("unset"),
+        agent_type = auth::get_optional_agent_type(&headers).as_deref().unwrap_or("unset"),
+        "Opening legacy SSE MCP stream"
+    );
     let mut broadcast_rx = state.event_tx.subscribe();
     let (session_tx, mut session_rx) = mpsc::channel(32);
     let session_id = Uuid::new_v4().to_string();
@@ -222,7 +338,7 @@ async fn sse_handler(
         }
     };
 
-    Sse::new(stream).keep_alive(KeepAlive::new().interval(Duration::from_secs(15)))
+    Ok(Sse::new(stream).keep_alive(KeepAlive::new().interval(Duration::from_secs(15))))
 }
 
 /// Message endpoint for JSON-RPC requests
@@ -232,7 +348,16 @@ async fn message_handler(
     headers: HeaderMap,
     Json(request): Json<JsonRpcRequest>,
 ) -> Response {
-    info!("Received MCP request: {}", request.method);
+    if let Err(status) = validate_origin(&headers) {
+        return status.into_response();
+    }
+
+    info!(
+        method = %request.method,
+        agent_id = auth::get_optional_agent_id(&headers).as_deref().unwrap_or("unset"),
+        agent_type = auth::get_optional_agent_type(&headers).as_deref().unwrap_or("unset"),
+        "Received legacy SSE MCP request"
+    );
 
     if let Some(session_id) = query.session_id.as_deref() {
         if !state.has_session(session_id).await {
@@ -513,4 +638,22 @@ mod tests {
             "/mcp/message?sessionId=abc123"
         );
     }
+
+    #[test]
+    fn allows_matching_origin_and_host() {
+        let mut headers = HeaderMap::new();
+        headers.insert(ORIGIN, "https://ob.ingwaz.work".parse().unwrap());
+        headers.insert(HOST, "ob.ingwaz.work".parse().unwrap());
+
+        assert!(validate_origin(&headers).is_ok());
+    }
+
+    #[test]
+    fn rejects_mismatched_origin_and_host() {
+        let mut headers = HeaderMap::new();
+        headers.insert(ORIGIN, "https://evil.example".parse().unwrap());
+        headers.insert(HOST, "ob.ingwaz.work".parse().unwrap());
+
+        assert_eq!(validate_origin(&headers), Err(StatusCode::FORBIDDEN));
+    }
 }

tests/e2e_mcp.rs

@@ -110,6 +110,26 @@ async fn call_jsonrpc(client: &reqwest::Client, base: &str, request: Value) -> V
         .expect("JSON-RPC response body")
 }
 
+async fn call_streamable_jsonrpc(
+    client: &reqwest::Client,
+    base: &str,
+    request: Value,
+) -> reqwest::Response {
+    let mut req_builder = client
+        .post(format!("{base}/mcp"))
+        .header("Accept", "application/json, text/event-stream")
+        .json(&request);
+
+    if let Some(key) = api_key() {
+        req_builder = req_builder.header("X-API-Key", key);
+    }
+
+    req_builder
+        .send()
+        .await
+        .expect("streamable JSON-RPC HTTP request")
+}
+
 /// Make an authenticated GET request to an MCP endpoint
 async fn get_mcp_endpoint(client: &reqwest::Client, base: &str, path: &str) -> reqwest::Response {
     let mut req_builder = client.get(format!("{base}{path}"));
@@ -357,6 +377,98 @@ async fn e2e_transport_tools_list_and_unknown_method() {
     );
 }
 
+#[tokio::test]
+async fn e2e_streamable_initialize_and_tools_list() {
+    let base = base_url();
+    let client = reqwest::Client::builder()
+        .timeout(Duration::from_secs(20))
+        .build()
+        .expect("reqwest client");
+
+    wait_until_ready(&client, &base).await;
+
+    let initialize_response: Value = call_streamable_jsonrpc(
+        &client,
+        &base,
+        json!({
+            "jsonrpc": "2.0",
+            "id": "streamable-init-1",
+            "method": "initialize",
+            "params": {
+                "protocolVersion": "2024-11-05",
+                "capabilities": {},
+                "clientInfo": {
+                    "name": "e2e-client",
+                    "version": "0.1.0"
+                }
+            }
+        }),
+    )
+    .await
+    .json()
+    .await
+    .expect("streamable initialize JSON");
+
+    assert_eq!(
+        initialize_response
+            .get("result")
+            .and_then(|value| value.get("protocolVersion"))
+            .and_then(Value::as_str),
+        Some("2024-11-05")
+    );
+
+    let tools_list_response: Value = call_streamable_jsonrpc(
+        &client,
+        &base,
+        json!({
+            "jsonrpc": "2.0",
+            "id": "streamable-tools-list-1",
+            "method": "tools/list",
+            "params": {}
+        }),
+    )
+    .await
+    .json()
+    .await
+    .expect("streamable tools/list JSON");
+
+    assert!(
+        tools_list_response
+            .get("result")
+            .and_then(|value| value.get("tools"))
+            .and_then(Value::as_array)
+            .map(|tools| !tools.is_empty())
+            .unwrap_or(false),
+        "streamable /mcp tools/list should return tool definitions"
+    );
+}
+
+#[tokio::test]
+async fn e2e_streamable_get_returns_405() {
+    let base = base_url();
+    let client = reqwest::Client::builder()
+        .timeout(Duration::from_secs(20))
+        .build()
+        .expect("reqwest client");
+
+    wait_until_ready(&client, &base).await;
+
+    let mut request = client
+        .get(format!("{base}/mcp"))
+        .header("Accept", "text/event-stream");
+
+    if let Some(key) = api_key() {
+        request = request.header("X-API-Key", key);
+    }
+
+    let response = request.send().await.expect("GET /mcp");
+    assert_eq!(
+        response.status(),
+        reqwest::StatusCode::METHOD_NOT_ALLOWED,
+        "streamable GET /mcp should explicitly return 405 when standalone SSE streams are not offered"
+    );
+}
+
 #[tokio::test]
 async fn e2e_purge_requires_confirm_flag() {
     let base = base_url();
@@ -879,60 +991,169 @@ async fn e2e_auth_enabled_accepts_test_key() {
 
 #[tokio::test]
 async fn e2e_batch_store_basic() -> anyhow::Result<()> {
-    let agent = format!("batch_{}", uuid::Uuid::new_v4());
-    let _ = db.purge_memories(&agent, None).await;
+    let base = base_url();
+    let client = reqwest::Client::builder()
+        .timeout(Duration::from_secs(20))
+        .build()
+        .expect("reqwest client");
 
-    let resp = client.call_tool("batch_store", serde_json::json!({
+    ensure_schema().await;
+    wait_until_ready(&client, &base).await;
+
+    let agent = format!("batch_{}", uuid::Uuid::new_v4());
+    let _ = call_tool(
+        &client,
+        &base,
+        "purge",
+        json!({ "agent_id": agent.clone(), "confirm": true }),
+    )
+    .await;
+
+    let result = call_tool(&client, &base, "batch_store", serde_json::json!({
         "agent_id": agent.clone(),
         "entries": [
             {"content": "Fact alpha for batch test"},
             {"content": "Fact beta for batch test"},
             {"content": "Fact gamma for batch test"}
         ]
-    })).await?;
+    })).await;
+
+    let _ = call_tool(
+        &client,
+        &base,
+        "purge",
+        json!({ "agent_id": agent, "confirm": true }),
+    )
+    .await;
 
-    let result: Value = serde_json::from_str(&resp.content[0].text)?;
     assert!(result["success"].as_bool().unwrap_or(false));
     assert_eq!(result["count"].as_i64().unwrap_or(0), 3);
-
-    db.purge_memories(&agent, None).await?;
     Ok(())
 }
 
 #[tokio::test]
 async fn e2e_batch_store_empty_rejected() -> anyhow::Result<()> {
-    let resp = client.call_tool("batch_store", serde_json::json!({
+    let base = base_url();
+    let client = reqwest::Client::builder()
+        .timeout(Duration::from_secs(20))
+        .build()
+        .expect("reqwest client");
+
+    wait_until_ready(&client, &base).await;
+
+    let response = call_jsonrpc(
+        &client,
+        &base,
+        json!({
+            "jsonrpc": "2.0",
+            "id": "batch-empty-1",
+            "method": "tools/call",
+            "params": {
+                "name": "batch_store",
+                "arguments": {
                     "entries": []
-    })).await;
-    assert!(resp.is_err() || resp.as_ref().unwrap().is_error());
+                }
+            }
+        }),
+    )
+    .await;
+
+    assert!(response.get("error").is_some(), "empty batch_store should return an error");
     Ok(())
 }
 
 #[tokio::test]
 async fn e2e_batch_store_exceeds_max() -> anyhow::Result<()> {
+    let base = base_url();
+    let client = reqwest::Client::builder()
+        .timeout(Duration::from_secs(20))
+        .build()
+        .expect("reqwest client");
+
+    wait_until_ready(&client, &base).await;
+
     let entries: Vec<Value> = (0..51).map(|i| serde_json::json!({"content": format!("Entry {}", i)})).collect();
-    let resp = client.call_tool("batch_store", serde_json::json!({
+    let response = call_jsonrpc(
+        &client,
+        &base,
+        json!({
+            "jsonrpc": "2.0",
+            "id": "batch-too-large-1",
+            "method": "tools/call",
+            "params": {
+                "name": "batch_store",
+                "arguments": {
                     "entries": entries
-    })).await;
-    assert!(resp.is_err() || resp.as_ref().unwrap().is_error());
+                }
+            }
+        }),
+    )
+    .await;
+
+    assert!(response.get("error").is_some(), "oversized batch_store should return an error");
     Ok(())
 }
 
 #[tokio::test]
 async fn e2e_batch_store_missing_content() -> anyhow::Result<()> {
-    let resp = client.call_tool("batch_store", serde_json::json!({
+    let base = base_url();
+    let client = reqwest::Client::builder()
+        .timeout(Duration::from_secs(20))
+        .build()
+        .expect("reqwest client");
+
+    wait_until_ready(&client, &base).await;
+
+    let response = call_jsonrpc(
+        &client,
+        &base,
+        json!({
+            "jsonrpc": "2.0",
+            "id": "batch-missing-content-1",
+            "method": "tools/call",
+            "params": {
+                "name": "batch_store",
+                "arguments": {
                     "entries": [{"content": "Valid entry"}, {"metadata": {}}]
-    })).await;
-    assert!(resp.is_err() || resp.as_ref().unwrap().is_error());
+                }
+            }
+        }),
+    )
+    .await;
+
+    assert!(response.get("error").is_some(), "missing batch entry content should return an error");
     Ok(())
 }
 
 #[tokio::test]
 async fn e2e_batch_store_appears_in_tools() -> anyhow::Result<()> {
-    let tools = client.list_tools().await?;
-    let parsed: Value = serde_json::from_str(&tools.content[0].text)?;
-    let names: Vec<&str> = parsed.as_array().unwrap().iter()
-        .filter_map(|t| t.get("name").and_then(|n| n.as_str()))
+    let base = base_url();
+    let client = reqwest::Client::builder()
+        .timeout(Duration::from_secs(20))
+        .build()
+        .expect("reqwest client");
+
+    wait_until_ready(&client, &base).await;
+
+    let response = call_jsonrpc(
+        &client,
+        &base,
+        json!({
+            "jsonrpc": "2.0",
+            "id": "batch-tools-list-1",
+            "method": "tools/list",
+            "params": {}
+        }),
+    )
+    .await;
+
+    let names: Vec<&str> = response
+        .get("result")
+        .and_then(|value| value.get("tools"))
+        .and_then(Value::as_array)
+        .expect("tools/list result.tools")
+        .iter()
+        .filter_map(|t| t.get("name").and_then(Value::as_str))
         .collect();
     assert!(names.contains(&"batch_store"));
     Ok(())
@@ -940,14 +1161,37 @@ async fn e2e_batch_store_appears_in_tools() -> anyhow::Result<()> {
 
 #[tokio::test]
 async fn e2e_existing_store_unchanged() -> anyhow::Result<()> {
+    let base = base_url();
+    let client = reqwest::Client::builder()
+        .timeout(Duration::from_secs(20))
+        .build()
+        .expect("reqwest client");
+
+    ensure_schema().await;
+    wait_until_ready(&client, &base).await;
+
     let agent = format!("compat_{}", uuid::Uuid::new_v4());
-    let _ = db.purge_memories(&agent, None).await;
-    let resp = client.call_tool("store", serde_json::json!({
+    let _ = call_tool(
+        &client,
+        &base,
+        "purge",
+        json!({ "agent_id": agent.clone(), "confirm": true }),
+    )
+    .await;
+
+    let result = call_tool(&client, &base, "store", serde_json::json!({
         "agent_id": agent.clone(),
         "content": "Original store still works"
-    })).await?;
-    let result: Value = serde_json::from_str(&resp.content[0].text)?;
+    })).await;
+
+    let _ = call_tool(
+        &client,
+        &base,
+        "purge",
+        json!({ "agent_id": agent, "confirm": true }),
+    )
+    .await;
+
     assert!(result["success"].as_bool().unwrap_or(false));
-    db.purge_memories(&agent, None).await?;
     Ok(())
 }