diff --git a/strix/tools/agents_graph/tools.py b/strix/tools/agents_graph/tools.py
index a72b311c3..28a8c622f 100644
--- a/strix/tools/agents_graph/tools.py
+++ b/strix/tools/agents_graph/tools.py
@@ -350,13 +350,13 @@ async def wait_for_message(  # noqa: PLR0911
     )
 
 
-@function_tool(timeout=120)
+@function_tool(timeout=120, strict_mode=False)
 async def create_agent(
     ctx: RunContextWrapper,
     name: str,
     task: str,
     inherit_context: bool = True,
-    skills: list[str] | None = None,
+    skills: str | list[str] | None = None,
 ) -> str:
     """Spawn a specialist child agent to run in parallel.
 
@@ -419,6 +419,23 @@ async def create_agent(
             default=str,
         )
 
+    # Tolerate LLM providers that pass array params as JSON-encoded strings.
+    # Validate decoded shape: reject anything that isn't a list of strings.
+    if isinstance(skills, str):
+        original_skills = skills
+        try:
+            decoded_skills = json.loads(skills)
+        except json.JSONDecodeError:
+            skills = [s.strip() for s in skills.split(",") if s.strip()]
+        else:
+            if isinstance(decoded_skills, str):
+                skills = [decoded_skills]
+            elif isinstance(decoded_skills, list) and all(
+                isinstance(skill, str) for skill in decoded_skills
+            ):
+                skills = decoded_skills
+            else:
+                skills = [original_skills]
     skill_list = list(skills or [])
     skill_error = validate_requested_skills(skill_list)
     if skill_error:
diff --git a/strix/tools/load_skill/tool.py b/strix/tools/load_skill/tool.py
index 3ef6f9ec0..b3c7e413e 100644
--- a/strix/tools/load_skill/tool.py
+++ b/strix/tools/load_skill/tool.py
@@ -2,13 +2,15 @@
 
 from __future__ import annotations
 
+import json
+
 from agents import RunContextWrapper, function_tool
 
 from strix.skills import load_skills, validate_requested_skills
 
 
-@function_tool(timeout=10)
-async def load_skill(ctx: RunContextWrapper, skills: list[str]) -> str:
+@function_tool(timeout=10, strict_mode=False)
+async def load_skill(ctx: RunContextWrapper, skills: str | list[str]) -> str:
     """Return the markdown body of one or more skills as reference material.
 
     Use this when you need exact syntax / workflow / payload guidance
@@ -25,6 +27,23 @@ async def load_skill(ctx: RunContextWrapper, skills: list[str]) -> str:
             ``strix/skills/<category>/<name>.md``.
     """
     del ctx
+    # Tolerate LLM providers that pass array params as JSON-encoded strings.
+    # Validate decoded shape: reject anything that isn't a list of strings.
+    if isinstance(skills, str):
+        original_skills = skills
+        try:
+            decoded_skills = json.loads(skills)
+        except json.JSONDecodeError:
+            skills = [s.strip() for s in skills.split(",") if s.strip()]
+        else:
+            if isinstance(decoded_skills, str):
+                skills = [decoded_skills]
+            elif isinstance(decoded_skills, list) and all(
+                isinstance(skill, str) for skill in decoded_skills
+            ):
+                skills = decoded_skills
+            else:
+                skills = [original_skills]
     requested = list(skills or [])
     err = validate_requested_skills(requested)
     if err: