fix: generate agents with platform LLM grants

agent_builder/skills/deepagents-implementation-patterns/SKILL.md

@@ -1,12 +1,14 @@
 ---
 name: deepagents-implementation-patterns
-description: Implement DeepAgents inside a2a-pack agents with real deterministic tools, project skills, custom subagents, workspace-backed files, and caller-provided ChatOpenAI models. Use when editing create_deep_agent calls or deciding what belongs in tools, skills, or subagents.
+description: Implement DeepAgents inside a2a-pack agents with real deterministic tools, project skills, custom subagents, workspace-backed files, and ctx.llm-backed ChatOpenAI models. Use when editing create_deep_agent calls or deciding what belongs in tools, skills, or subagents.
 ---
 # DeepAgents Implementation Patterns
 
 DeepAgents supplies the inner agent loop. In this platform, the reliable shape
-is: caller-provided `ChatOpenAI`, A2A workspace backend, project DeepAgents
-skills, and a small number of exact tools.
+is: `ctx.llm`-backed `ChatOpenAI`, A2A workspace backend, project DeepAgents
+skills, and a small number of exact tools. Hosted generated agents should
+declare `LLMProvisioning.PLATFORM`; explicit BYOK agents can declare
+`CALLER_PROVIDED`, but both modes still read the same `ctx.llm` object.
 
 ## Implementation Shape
 
@@ -32,6 +34,11 @@ stable output paths only when that is one of the grant write prefixes.
 
 def build_graph(ctx: Any) -> Any:
     creds = ctx.llm
+    if not creds.api_key:
+        raise RuntimeError(
+            "LLM credentials were not available; declare PLATFORM for hosted "
+            "generated agents or configure caller-provided credentials."
+        )
     model = ChatOpenAI(
         model=creds.model,
         base_url=creds.base_url,