tinyagentos/routes/cluster.py (2)

19-33: ⚠️ Potential issue | 🟠 Major | ⚡ Quick win

Fix mutable default values in WorkerRegister (shared state across requests).

WorkerRegister defines mutable defaults:

• hardware: dict = {} (Line 22)
• backends: list[dict] = [] (Line 23)
• models: list[str] = [] (Line 24)
• capabilities: list[str] = [] (Line 25)

Even if Pydantic often copies defaults, relying on that is risky: mutable defaults can lead to cross-request data leakage / unexpected mutation when any code mutates these values.

Suggested fix: switch to Field(default_factory=...) for each mutable field.

🛠️ Proposed fix

-from pydantic import BaseModel
+from pydantic import BaseModel, Field

 class WorkerRegister(BaseModel):
     name: str
     url: str
-    hardware: dict = {}
-    backends: list[dict] = []
-    models: list[str] = []
-    capabilities: list[str] = []
+    hardware: dict = Field(default_factory=dict)
+    backends: list[dict] = Field(default_factory=list)
+    models: list[str] = Field(default_factory=list)
+    capabilities: list[str] = Field(default_factory=list)
     platform: str = ""

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@tinyagentos/routes/cluster.py` around lines 19 - 33, WorkerRegister uses
mutable default values for hardware, backends, models, and capabilities which
can cause shared state; update these fields to use Pydantic's Field with
default_factory (e.g., Field(default_factory=dict) for hardware and
Field(default_factory=list) for backends/models/capabilities) and import Field
from pydantic, leaving other fields and kv_* defaults unchanged so each model
instance gets its own fresh collection.

---

66-89: ⚠️ Potential issue | 🟠 Major

API contract mismatch: list_workers returns bare array, but consumer expects wrapped object.

The endpoint returns a bare JSON array, but tinyagentos/scheduling/resource_manager.py (line 112) calls .get("workers", []) expecting an object shape {"workers": [...]}. When the actual list response is received, .get() fails silently due to exception handling and returns [], causing cluster workers to never be discovered during remote resource checks.

Fix: In resource_manager.py line 112, change resp.json().get("workers", []) to resp.json() to match the actual endpoint response.

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@tinyagentos/routes/cluster.py` around lines 66 - 89, The remote workers API
(list_workers) returns a bare JSON array but the client is calling
resp.json().get("workers", []), causing discovery to fail; locate the call that
uses resp.json().get("workers", []) (the code that fetches remote cluster
workers) and replace that expression with resp.json() so the client treats the
response as the list itself (ensure the returned value is assigned to the
variable used as the workers list).