feat(skills): skills impacts /tools

Author: jrobertboos

src/app/endpoints/tools.py

@@ -28,6 +28,7 @@
     mcp_headers_dependency,
 )
 from utils.mcp_oauth_probe import check_mcp_auth
+from utils.pydantic_ai import get_agent_capability_tools
 from utils.tool_formatter import format_tools_list
 
 logger = get_logger(__name__)
@@ -238,11 +239,25 @@ async def tools_endpoint_handler(  # pylint: disable=too-many-locals,too-many-st
             server_source,
         )
 
+    existing_tool_ids = {
+        tool.get("identifier") for tool in consolidated_tools if tool.get("identifier")
+    }
+    capability_tools = get_agent_capability_tools(configuration.skills)
+    for tool_dict in capability_tools:
+        identifier = tool_dict.get("identifier")
+        if identifier and identifier not in existing_tool_ids:
+            consolidated_tools.append(tool_dict)
+            existing_tool_ids.add(identifier)
+
+    builtin_tool_count = len(
+        [t for t in consolidated_tools if t.get("server_source") == "builtin"]
+    )
+    mcp_tool_count = len(consolidated_tools) - builtin_tool_count
     logger.info(
         "Retrieved total of %d tools (%d from built-in toolgroups, %d from MCP servers)",
         len(consolidated_tools),
-        len([t for t in consolidated_tools if t.get("server_source") == "builtin"]),
-        len([t for t in consolidated_tools if t.get("server_source") != "builtin"]),
+        builtin_tool_count,
+        mcp_tool_count,
     )
 
     # Format tools with structured description parsing

src/utils/pydantic_ai.py

@@ -2,6 +2,7 @@
 
 from __future__ import annotations
 
+import re
 from typing import Any, Final, Optional, cast
 
 from llama_stack.core.library_client import AsyncLlamaStackAsLibraryClient
@@ -18,6 +19,11 @@
     LlamaStackResponsesModel,
 )
 
+_AGENT_SKILLS_PROVIDER_ID: Final[str] = "agent-skills"
+_AGENT_SKILLS_TOOLGROUP_ID: Final[str] = "builtin::agent-skills"
+_BUILTIN_CAPABILITY_SERVER_SOURCE: Final[str] = "builtin"
+_CAPABILITY_TOOL_TYPE: Final[str] = "tool"
+
 _LLS_RESPONSES_EXTRA_FIELDS: Final[frozenset[str]] = frozenset(
     {
         "conversation",
@@ -96,6 +102,92 @@ def _skills_capability(
     )
 
 
+def _json_schema_to_parameters(
+    schema: Optional[dict[str, Any]],
+) -> list[dict[str, Any]]:
+    """Convert a JSON Schema object to the flat parameter list used by ``/tools``."""
+    if not schema or "properties" not in schema:
+        return []
+
+    required_params = set(schema.get("required", []))
+    parameters: list[dict[str, Any]] = []
+    for name, prop in schema["properties"].items():
+        parameter_type = prop.get("type")
+        if parameter_type is None and "anyOf" in prop:
+            for option in prop["anyOf"]:
+                if isinstance(option, dict) and option.get("type") not in (
+                    None,
+                    "null",
+                ):
+                    parameter_type = option["type"]
+                    break
+        parameters.append(
+            {
+                "name": name,
+                "description": prop.get("description", ""),
+                "parameter_type": parameter_type or "string",
+                "required": name in required_params,
+                "default": prop.get("default"),
+            }
+        )
+    return parameters
+
+
+def _capability_tool_description(description: str) -> str:
+    """Extract a user-facing description from pydantic-ai tool docstrings."""
+    if match := re.search(r"<summary>(.*?)</summary>", description, re.DOTALL):
+        return match.group(1).strip()
+    return description.strip()
+
+
+def _capability_tools_from_toolset(toolset: Any) -> list[dict[str, Any]]:
+    """Serialize tools registered on a pydantic-ai capability toolset."""
+    raw_tools = getattr(toolset, "tools", None)
+    if not raw_tools:
+        return []
+
+    tool_dicts: list[dict[str, Any]] = []
+    for tool in raw_tools.values():
+        tool_dicts.append(
+            {
+                "identifier": tool.name,
+                "description": _capability_tool_description(tool.description or ""),
+                "parameters": _json_schema_to_parameters(
+                    tool.function_schema.json_schema
+                ),
+                "provider_id": _AGENT_SKILLS_PROVIDER_ID,
+                "toolgroup_id": _AGENT_SKILLS_TOOLGROUP_ID,
+                "server_source": _BUILTIN_CAPABILITY_SERVER_SOURCE,
+                "type": _CAPABILITY_TOOL_TYPE,
+            }
+        )
+    return tool_dicts
+
+
+def get_agent_capability_tools(
+    skills: Optional[SkillsConfiguration],
+) -> list[dict[str, Any]]:
+    """Return tool metadata for pydantic-ai capabilities configured for LCS agents.
+
+    Parameters:
+        skills: Agent skills configuration from LCS, or None when skills are disabled.
+
+    Returns:
+        Tool dictionaries compatible with the ``/tools`` endpoint response format.
+    """
+    capabilities = _agent_capabilities(skills) or []
+
+    tools: list[dict[str, Any]] = []
+    for capability in capabilities:
+        if not isinstance(capability, AbstractCapability):
+            continue
+        toolset = capability.get_toolset()
+        if toolset is None:
+            continue
+        tools.extend(_capability_tools_from_toolset(toolset))
+    return tools
+
+
 def _agent_capabilities(
     skills: Optional[SkillsConfiguration],
     no_tools: bool = False,

tests/unit/app/endpoints/test_tools.py

@@ -23,6 +23,7 @@
     LlamaStackConfiguration,
     ModelContextProtocolServer,
     ServiceConfiguration,
+    SkillsConfiguration,
     TLSConfiguration,
     UserDataCollection,
 )
@@ -1135,3 +1136,44 @@ async def test_tools_endpoint_empty_legacy_fields_overridden(
     assert tool["parameters"][0]["name"] == "query"
     assert tool["parameters"][0]["parameter_type"] == "string"
     assert tool["parameters"][0]["required"] is True
+
+
+@pytest.mark.asyncio
+async def test_tools_endpoint_includes_agent_capability_tools(
+    mocker: MockerFixture,
+    mock_configuration: Configuration,  # pylint: disable=redefined-outer-name
+    mock_skills_configuration: SkillsConfiguration,
+) -> None:
+    """Test that configured pydantic-ai capabilities appear in /tools output."""
+    config_with_skills = mock_configuration.model_copy(
+        update={"skills": mock_skills_configuration}
+    )
+    app_config = AppConfig()
+    app_config._configuration = config_with_skills
+    mocker.patch("app.endpoints.tools.configuration", app_config)
+    mocker.patch("app.endpoints.tools.authorize", lambda _: lambda func: func)
+
+    mock_client_holder = mocker.patch("app.endpoints.tools.AsyncLlamaStackClientHolder")
+    mock_client = mocker.AsyncMock()
+    mock_client_holder.return_value.get_client.return_value = mock_client
+    mock_client.toolgroups.list.return_value = []
+
+    mock_request = mocker.Mock()
+    mock_auth = MOCK_AUTH
+
+    response = await tools.tools_endpoint_handler.__wrapped__(  # pyright: ignore
+        mock_request, mock_auth, {}
+    )
+
+    tool_ids = [tool["identifier"] for tool in response.tools]
+    assert "list_skills" in tool_ids
+    assert "load_skill" in tool_ids
+    assert "read_skill_resource" in tool_ids
+    assert "run_skill_script" in tool_ids
+
+    list_skills = next(
+        tool for tool in response.tools if tool["identifier"] == "list_skills"
+    )
+    assert list_skills["provider_id"] == "agent-skills"
+    assert list_skills["toolgroup_id"] == "builtin::agent-skills"
+    assert list_skills["server_source"] == "builtin"

tests/unit/utils/test_pydantic_ai.py

@@ -17,6 +17,7 @@
     _model_settings_from_responses_params,
     _skills_capability,
     build_agent,
+    get_agent_capability_tools,
     llama_stack_provider_from_client,
 )
 
@@ -363,3 +364,46 @@ def test_agent_excludes_tool_capabilities_when_no_tools(
             type(capability) for capability in agent._root_capability.capabilities
         }
         assert SkillsCapability not in capability_types
+
+
+class TestGetAgentCapabilityTools:
+    """Tests for get_agent_capability_tools."""
+
+    def test_returns_empty_list_when_skills_not_configured(self) -> None:
+        """Test that missing skills configuration yields no capability tools."""
+        assert not get_agent_capability_tools(None)
+        assert not get_agent_capability_tools(SkillsConfiguration(paths=[]))
+
+    def test_returns_skills_tools_when_configured(
+        self, mock_skills_configuration: SkillsConfiguration
+    ) -> None:
+        """Test that configured skills expose pydantic-ai skill tools."""
+        tools = get_agent_capability_tools(mock_skills_configuration)
+
+        assert [tool["identifier"] for tool in tools] == [
+            "list_skills",
+            "load_skill",
+            "read_skill_resource",
+            "run_skill_script",
+        ]
+        assert all(
+            tool["provider_id"] == "agent-skills"
+            and tool["toolgroup_id"] == "builtin::agent-skills"
+            and tool["server_source"] == "builtin"
+            and tool["type"] == "tool"
+            for tool in tools
+        )
+
+        load_skill = next(tool for tool in tools if tool["identifier"] == "load_skill")
+        assert load_skill["parameters"] == [
+            {
+                "name": "skill_name",
+                "description": (
+                    "Exact name from your available skills list.\n"
+                    'Must match exactly (e.g., "data-analysis" not "data analysis").'
+                ),
+                "parameter_type": "string",
+                "required": True,
+                "default": None,
+            }
+        ]