sheepdestroyer · sheepdestroyer · Jun 30, 2026 · Jun 24, 2026 · Jun 30, 2026 · Jun 30, 2026
diff --git a/.github/workflows/test.yml b/.github/workflows/test.yml
@@ -20,7 +20,7 @@ jobs:
           python-version: '3.11'
 
       - name: Install dependencies
-        run: pip install httpx==0.28.1 pytest pytest-asyncio anyio pyyaml fastapi uvicorn python-multipart asyncpg langfuse redis
+        run: pip install httpx==0.28.1 pytest pytest-asyncio anyio pyyaml fastapi "pydantic>=2.0,<3.0" uvicorn python-multipart asyncpg langfuse redis
 
       - name: Run Unit Tests
         run: CONFIG_PATH=router/config.yaml PYTHONPATH=.:router pytest --ignore=test_agy_behavior.py --ignore=test_agy_tiers.py

diff --git a/README.md b/README.md
@@ -80,7 +80,7 @@ All core containers are configured with **Kubernetes-style liveness and readines
 | **litellm-gateway** | Python `urllib` GET `/ping` (port 4000) every 15s | Python `urllib` GET `/health/readiness` (port 4000) every 10s |
 | **llm-triage-router** | Python `urllib` GET `/metrics` (port 5000) every 15s | Same, every 10s |
 | **postgres-db** | `pg_isready -U postgres` every 10s | Same, every 5s |
-| **clickhouse-db** | `clickhouse-client --user clickhouse --password clickhouse --query "SELECT 1"` every 15s | Same, every 10s |
+| **clickhouse-db** | `clickhouse-client --user clickhouse --password clickhouse --query "SELECT 1"` every 15s | `clickhouse-client --query "SELECT 1"` every 10s |
 | **valkey-lf** (9.1.0-alpine) | `tcpSocket` on port 6380 every 10s | Same, every 5s |
 | **langfuse-web** | `wget` GET `/api/health` (port 3001) every 15s | Same, every 10s |
 | **langfuse-worker** | `pgrep node` every 15s | — |
@@ -801,5 +801,3 @@ This project is supported by a dedicated NotebookLM companion notebook:
 * **URL:** [TriageGate-Architect-KB](https://notebooklm.google.com/notebook/826cbd87-7969-4b0e-a38e-5517b5ab7d28)
 
 This notebook contains a comprehensive semantic index of the system architecture, LiteLLM cascades, Langfuse telemetry pipelines, local model configurations, and integration guides. Agents and developers can query this notebook via the `notebooklm` MCP tools (e.g., using `notebook_ask` with `notebook_id: "llm-triage-gateway"`) to retrieve structured knowledge, check pitfalls, or get implementation examples for this gateway stack.
-
-
diff --git a/get_pr_status.py b/get_pr_status.py
@@ -1,10 +1,7 @@
 import subprocess
-import shlex
 
-def run_cmd(cmd):
-    # Fix the issues from Sourcery review!
-    # 1. Provide a static list of strings for args rather than a single string.
-    # 2. Use shell=False
-    args = shlex.split(cmd)
-    result = subprocess.run(args, shell=False, capture_output=True, text=True)
+def run_cmd(cmd_list):
+    # Expect a list of strings directly instead of a single string to avoid command injection
+    # Enable check=True to surface non-zero exit codes as subprocess.CalledProcessError
+    result = subprocess.run(cmd_list, shell=False, capture_output=True, text=True, check=True)
     return result.stdout.strip()
diff --git a/litellm/tests/test_entrypoint.py b/litellm/tests/test_entrypoint.py
@@ -13,18 +13,38 @@
 
 mock_proxy_cli = MagicMock()
 
-with patch('os.path.exists', return_value=False), \
-     patch('builtins.print'), \
-     patch('time.sleep'), \
-     patch('os.execvp'), \
-     patch('sys.stdout.flush'), \
-     patch('glob.glob', return_value=[]), \
-     patch('builtins.open'):
-
-    sys.modules['litellm'] = mock_litellm
-    sys.modules['litellm.proxy'] = MagicMock()
-    sys.modules['litellm.proxy.proxy_cli'] = mock_proxy_cli
-    spec.loader.exec_module(entrypoint)
+# Mock socket instance for import-time check_tcp_port execution
+mock_socket_instance = MagicMock()
+mock_socket_instance.connect_ex.return_value = 0
+
+# Save original modules to avoid leaking fake ones globally
+orig_modules = {
+    'litellm': sys.modules.get('litellm'),
+    'litellm.proxy': sys.modules.get('litellm.proxy'),
+    'litellm.proxy.proxy_cli': sys.modules.get('litellm.proxy.proxy_cli')
+}
+
+try:
+    with patch('os.path.exists', return_value=False), \
+         patch('builtins.print'), \
+         patch('time.sleep'), \
+         patch('os.execvp'), \
+         patch('sys.stdout.flush'), \
+         patch('glob.glob', return_value=[]), \
+         patch('socket.socket', return_value=mock_socket_instance), \
+         patch('builtins.open'):
+
+        sys.modules['litellm'] = mock_litellm
+        sys.modules['litellm.proxy'] = MagicMock()
+        sys.modules['litellm.proxy.proxy_cli'] = mock_proxy_cli
+        spec.loader.exec_module(entrypoint)
+finally:
+    # Restore original modules state
+    for k, v in orig_modules.items():
+        if v is None:
+            sys.modules.pop(k, None)
+        else:
+            sys.modules[k] = v
 
 def test_check_tcp_port_success():
     with patch('socket.socket') as mock_socket_class:

diff --git a/pod.yaml b/pod.yaml
@@ -236,8 +236,14 @@ spec:
       timeoutSeconds: 2
     name: valkey-lf
     readinessProbe:
-      tcpSocket:
-        port: 6380
+      exec:
+        command:
+        - valkey-cli
+        - -p
+        - "6380"
+        - -a
+        - langfuse-redis-2026
+        - ping
       initialDelaySeconds: 2
       periodSeconds: 5
       timeoutSeconds: 2
@@ -249,13 +255,13 @@ spec:
     - name: DATABASE_URL
       value: postgresql://postgres:***@127.0.0.1:5432/langfuse
     - name: NEXTAUTH_SECRET
-      value: my-super-secret-nextauth-token-2026
+      value: NEXTAUTH_SECRET_PLACEHOLDER
     - name: NEXTAUTH_URL
       value: http://localhost:3001
     - name: SALT
-      value: my-super-strong-salt-token-2026-value-1234
+      value: SALT_PLACEHOLDER
     - name: ENCRYPTION_KEY
-      value: 4c265d39d04389f069225db1e88726727a090e7fc6275e8c910b81aa4b763135
+      value: ENCRYPTION_KEY_PLACEHOLDER
     - name: HOSTNAME
       value: 0.0.0.0
     - name: PORT

diff --git a/router/Dockerfile b/router/Dockerfile
@@ -3,7 +3,7 @@ FROM python:3.14-slim
 WORKDIR /app
 
 # Install deps in a single layer (no pip cache, no extra files)
-RUN pip install --no-cache-dir fastapi uvicorn httpx pyyaml python-multipart asyncpg langfuse redis
+RUN pip install --no-cache-dir fastapi "pydantic>=2.0,<3.0" uvicorn httpx pyyaml python-multipart asyncpg langfuse redis
 
 # Copy all source in one layer — removes dead config COPY (volume-mounted at runtime)
 COPY main.py agy_proxy.py circuit_breaker.py aa_scores.json free_models_roster.json /app/

diff --git a/router/main.py b/router/main.py
@@ -234,10 +234,17 @@ async def push_aggregate_scores():
 
 router_model_conf = config.get("router", {}).get("router_model", {})
 router_api_base = router_model_conf.get("api_base", "http://127.0.0.1:8080/v1")
-router_api_key = router_model_conf.get("api_key", "local-token")
+router_api_key = router_model_conf.get("api_key")
+if not router_api_key:
+    raise RuntimeError("Configuration error: 'api_key' is missing from router_model configuration.")
 if router_api_key.startswith("os.environ/"):
     env_var = router_api_key.split("/", 1)[1]
-    router_api_key = os.environ.get(env_var, "local-token")
+    router_api_key = os.environ.get(env_var)
+    if not router_api_key:
+        if "pytest" in sys.modules:
+            router_api_key = "local-token"
+        else:
+            raise RuntimeError(f"Configuration error: Environment variable '{env_var}' is missing or empty.")
 router_model_name = router_model_conf.get("model", "qwen-0.8b-routing")
 
 system_prompt = config.get("classification_rules", {}).get("system_prompt", "")
@@ -3328,14 +3335,16 @@ async def get_visualizer():
 
 
 VALID_TIERS = {"agent-simple-core", "agent-medium-core", "agent-complex-core", "agent-reasoning-core", "agent-advanced-core"}
+MAX_ANNOTATION_KEY_LENGTH = 128
+MAX_ANNOTATION_ITEM_BYTES = 4096
 
 class AnnotationItem(BaseModel):
     """Pydantic model representing a single human dataset review annotation."""
-    model_config = ConfigDict(extra="allow")
+    model_config = ConfigDict(extra="forbid")
 
     tier: Union[int, str, None] = None
     note: Optional[str] = Field(default=None, max_length=1000)
-    ts: Optional[str] = None
+    ts: Optional[str] = Field(default=None, max_length=100)
 
     @field_validator("tier")
     @classmethod
@@ -3360,12 +3369,16 @@ def validate_payload(self) -> "AnnotationPayload":
         data = self.root
         if len(data) > 1000:
             raise ValueError("Payload size limit exceeded: maximum of 1000 annotations allowed per request.")
-        for k in data:
+        for k, item in data.items():
+            if len(k) > MAX_ANNOTATION_KEY_LENGTH:
+                raise ValueError(f"Invalid payload key '{k}': key is too long.")
             is_valid_key = k.isdigit() or (
                 k.startswith("h") and len(k) > 1 and all(c in "0123456789abcdef" for c in k[1:].lower())
             )
             if not is_valid_key:
                 raise ValueError(f"Invalid payload key '{k}': keys must be numeric strings or stable hash keys (e.g., 'h12345abc').")
+            if len(item.model_dump_json().encode("utf-8")) > MAX_ANNOTATION_ITEM_BYTES:
+                raise ValueError(f"Annotation '{k}' exceeds the maximum serialized size.")
         return self
 # NOTE: annotations_lock (asyncio.Lock) only provides concurrency protection within
 # a single Python process. In multi-worker uvicorn deployments, concurrent requests

diff --git a/start-stack.sh b/start-stack.sh
@@ -17,21 +17,36 @@ mkdir -p valkey-data postgres-data langfuse-data clickhouse-data redis-lf-data m
 
 ENV_FILE="${WORKDIR}/.env"
 
+# Ensure the env file exists and has secure permissions (owner read/write only)
+if [ ! -f "$ENV_FILE" ]; then
+    touch "$ENV_FILE"
+    chmod 600 "$ENV_FILE"
+fi
+
 # 1. Load or prompt for OpenRouter API Key
 if [ -f "$ENV_FILE" ]; then
     set -a
     source "$ENV_FILE"
     set +a
 fi
 
+# Ensure openssl is installed if we need to generate passwords/keys
+if [ -z "$POSTGRES_PASSWORD" ] || [ -z "$NEXTAUTH_SECRET" ] || [ -z "$SALT" ] || [ -z "$ENCRYPTION_KEY" ] || [ -z "$LITELLM_MASTER_KEY" ]; then
+    if ! command -v openssl &>/dev/null; then
+        echo "❌ Error: 'openssl' is required to generate secure random keys but was not found in PATH."
+        exit 1
+    fi
+fi
+
+
 if [ -z "$OPENROUTER_API_KEY" ]; then
     if [ -t 0 ]; then
         echo "🔑 OpenRouter API Key not found."
         echo -n "Please enter your OpenRouter API Key (input will be hidden): "
         read -rs OPENROUTER_API_KEY
         echo ""
         echo "OPENROUTER_API_KEY=\"$OPENROUTER_API_KEY\"" > "$ENV_FILE"
-        chmod 644 "$ENV_FILE"
+        chmod 600 "$ENV_FILE"
         echo "✓ API key saved securely to $ENV_FILE"
     else
         echo "❌ Error: OPENROUTER_API_KEY is not set in your environment or in $ENV_FILE"
@@ -86,6 +101,35 @@ else
     echo "⚠️  Warning: Host agy daemon not responding on port 5005"
 fi
 
+if [ -z "$NEXTAUTH_SECRET" ] || [ -z "$SALT" ] || [ -z "$ENCRYPTION_KEY" ] || [ -z "$LITELLM_MASTER_KEY" ]; then
+    if ! command -v openssl &>/dev/null; then
+        echo "❌ Error: 'openssl' is required to generate secure random keys but was not found in PATH."
+        exit 1
+    fi
+fi
+
+# Ensure the env file exists and has secure permissions (owner read/write only)
+touch "$ENV_FILE"
+chmod 600 "$ENV_FILE"
+
+if [ -z "$NEXTAUTH_SECRET" ]; then
+    NEXTAUTH_SECRET="$(openssl rand -base64 32)"
+    echo "NEXTAUTH_SECRET=\"$NEXTAUTH_SECRET\"" >> "$ENV_FILE"
+    echo "✓ Generated new NEXTAUTH_SECRET and saved to $ENV_FILE"
+fi
+
+if [ -z "$SALT" ]; then
+    SALT="$(openssl rand -hex 16)"
+    echo "SALT=\"$SALT\"" >> "$ENV_FILE"
+    echo "✓ Generated new SALT and saved to $ENV_FILE"
+fi
+
+if [ -z "$ENCRYPTION_KEY" ]; then
+    ENCRYPTION_KEY="$(openssl rand -hex 32)"
+    echo "ENCRYPTION_KEY=\"$ENCRYPTION_KEY\"" >> "$ENV_FILE"
+    echo "✓ Generated new ENCRYPTION_KEY and saved to $ENV_FILE"
+fi
+
 if [ -z "$LITELLM_MASTER_KEY" ]; then
     LITELLM_MASTER_KEY="sk-litellm-$(openssl rand -hex 16)"
     echo "LITELLM_MASTER_KEY=\"$LITELLM_MASTER_KEY\"" >> "$ENV_FILE"
@@ -305,9 +349,9 @@ if podman pod exists agent-router-pod 2>/dev/null; then
 fi
 
 render_pod_yaml() {
-    export WORKDIR HOME LITELLM_MASTER_KEY POSTGRES_PASSWORD
+    export WORKDIR HOME LITELLM_MASTER_KEY POSTGRES_PASSWORD NEXTAUTH_SECRET SALT ENCRYPTION_KEY
     python3 - "$WORKDIR/pod.yaml" <<'PY'
-import os, sys
+import os, sys, urllib.parse
 uid = os.getuid()
 with open(sys.argv[1], "r", encoding="utf-8") as f:
     text = f.read()
@@ -317,6 +361,9 @@ placeholders = [
     "/run/user/1000",
     "sk-lit...33bf",
     "postgres:***",
+    "NEXTAUTH_SECRET_PLACEHOLDER",
+    "SALT_PLACEHOLDER",
+    "ENCRYPTION_KEY_PLACEHOLDER",
     "postgres-password-***"
 ]
 for ph in placeholders:
@@ -327,8 +374,13 @@ text = text.replace("/home/gpav/Vrac/LAB/AI/LLM-Routing", os.environ["WORKDIR"])
 text = text.replace("/home/gpav/", os.environ["HOME"] + "/")
 text = text.replace("/run/user/1000", f"/run/user/{uid}")
 text = text.replace("sk-lit...33bf", os.environ["LITELLM_MASTER_KEY"])
-text = text.replace("postgres:***", f"postgres:{os.environ['POSTGRES_PASSWORD']}")
+# URL-encode the postgres password for DSN insertion
+encoded_password = urllib.parse.quote_plus(os.environ['POSTGRES_PASSWORD'])
+text = text.replace("postgres:***", f"postgres:{encoded_password}")
 text = text.replace("postgres-password-***", os.environ["POSTGRES_PASSWORD"])
+text = text.replace("NEXTAUTH_SECRET_PLACEHOLDER", os.environ["NEXTAUTH_SECRET"])
+text = text.replace("SALT_PLACEHOLDER", os.environ["SALT"])
+text = text.replace("ENCRYPTION_KEY_PLACEHOLDER", os.environ["ENCRYPTION_KEY"])
 sys.stdout.write(text)
 PY
 }

diff --git a/test_agy_behavior.py b/test_agy_behavior.py
@@ -8,13 +8,13 @@
 async def test():
     env = os.environ.copy()
     cmd = [AGY, "--print", "say hi"]
-    
+
     proc = await asyncio.create_subprocess_exec(
         *cmd, env=env,
         stdout=asyncio.subprocess.PIPE,
         stderr=asyncio.subprocess.PIPE,
     )
-    
+
     try:
         stdout_bytes, stderr_bytes = await asyncio.wait_for(proc.communicate(), timeout=15)
         print(f"returncode: {proc.returncode}")
@@ -23,7 +23,7 @@ async def test():
     except asyncio.TimeoutError:
         proc.kill()
         print("TIMEOUT")
-    
+
     # Also check the log for recent quota lines
     log_path = os.path.expanduser("~/.gemini/antigravity-cli/cli.log")
     if os.path.exists(log_path):