data-science-mcp 0.2.0__py3-none-any.whl

data_science_mcp/__init__.py

@@ -0,0 +1,70 @@
+#!/usr/bin/env python
+# coding: utf-8
+
+import importlib
+import inspect
+import warnings
+from typing import List
+
+# Suppress RequestsDependencyWarning due to chardet 6.x / requests 2.32.x mismatch
+# Centralized here to ensure it runs before any sub-package imports
+warnings.filterwarnings("ignore", message=".*urllib3.*or chardet.*")
+
+__all__: List[str] = []
+
+CORE_MODULES = [
+    "data_science_mcp.api_client",
+]
+
+OPTIONAL_MODULES = {
+    "data_science_mcp.agent_server": "agent",
+    "data_science_mcp.mcp_server": "mcp",
+}
+
+
+def _import_module_safely(module_name: str):
+    """Try to import a module and return it, or None if not available."""
+    try:
+        return importlib.import_module(module_name)
+    except ImportError:
+        return None
+
+
+def _expose_members(module):
+    """Expose public classes and functions from a module into globals and __all__."""
+    for name, obj in inspect.getmembers(module):
+        if (inspect.isclass(obj) or inspect.isfunction(obj)) and not name.startswith(
+            "_"
+        ):
+            globals()[name] = obj
+            __all__.append(name)
+
+
+for module_name in CORE_MODULES:
+    try:
+        module = importlib.import_module(module_name)
+        _expose_members(module)
+    except ImportError:
+        pass
+
+for module_name, extra_name in OPTIONAL_MODULES.items():
+    module = _import_module_safely(module_name)
+    if module is not None:
+        _expose_members(module)
+        globals()[f"_{extra_name.upper()}_AVAILABLE"] = True
+    else:
+        globals()[f"_{extra_name.upper()}_AVAILABLE"] = False
+
+_MCP_AVAILABLE = OPTIONAL_MODULES.get("data_science_mcp.mcp_server") in [
+    m.__name__ for m in globals().values() if hasattr(m, "__name__")
+]
+_AGENT_AVAILABLE = "data_science_mcp.agent_server" in globals()
+
+__all__.extend(["_MCP_AVAILABLE", "_AGENT_AVAILABLE"])
+
+
+"""
+data-science-mcp
+
+Data Science MCP Server — Model training, evaluation, and evolution tools for agentic ML workflows. Integrates with agent-utilities IModelEvolver (CONCEPT:AHE-3.15).
+"""

data_science_mcp/__main__.py

@@ -0,0 +1,6 @@
+#!/usr/bin/python
+# coding: utf-8
+from data_science_mcp.agent_server import agent_server
+
+if __name__ == "__main__":
+    agent_server()

data_science_mcp/agent/AGENTS.md

@@ -0,0 +1,12 @@
+# AGENTS.md - Known A2A Peer Agents
+Last updated: 2026-05-06
+
+This file is the local registry of other A2A agents this agent can discover and call.
+
+## Registered A2A Peers
+
+| Name            | Endpoint URL                    | Description                          | Capabilities                     | Auth      | Notes / Last Connected |
+|-----------------|---------------------------------|--------------------------------------|----------------------------------|-----------|------------------------|
+| SearchMaster    | http://search-agent:9000/a2a    | Advanced web researcher              | web_search, summarize, browse    | none      | 2026-05-06             |
+
+*Add new rows manually or let the agent call `register_a2a_peer(...)`.*

data_science_mcp/agent/CRON.md

@@ -0,0 +1,12 @@
+# CRON.md - Persistent Scheduled Tasks
+Last updated: 2026-05-06
+
+## Active Tasks
+
+| ID          | Name              | Interval (min) | Prompt                              | Last run          | Next approx |
+|-------------|-------------------|----------------|-------------------------------------|-------------------|-------------|
+| heartbeat   | Heartbeat         | 30             | @HEARTBEAT.md                       | —                 | —           |
+| log-cleanup | Log Cleanup       | 720            | __internal:cleanup_cron_log         | —                 | —           |
+
+*Edit this table to add/remove tasks. The agent reloads it periodically.*
+*Use `@filename.md` in the Prompt column to load a multi-line prompt from a workspace file.*

data_science_mcp/agent/CRON_LOG.md

@@ -0,0 +1,5 @@
+# CRON_LOG.md - Scheduled Task History
+Last updated: 2026-05-06
+
+| Timestamp | Task ID | Status | Message |
+|-----------|---------|--------|---------|

data_science_mcp/agent/HEARTBEAT.md

@@ -0,0 +1,28 @@
+# Heartbeat — Periodic Self-Check
+
+You are running a scheduled heartbeat. Perform these checks and report results concisely.
+
+## Checks
+
+1. **Tool Availability** — Call `list_tools` or equivalent to verify your MCP tools are reachable. Report any connection failures.
+2. **Memory Review** — Query the **Knowledge Graph** for any pending follow-up tasks, architectural decisions, or action items.
+3. **Cron Log** — Read `CRON_LOG.md` and check for recent errors (❌). Summarize any failures from the last 24 hours.
+4. **Peer Agents** — Read `AGENTS.md` and note if any registered peers need attention.
+5. **Domain-Specific Checks**:
+   - **Service Health**: Check service health status and scan recent logs for critical errors using available tools.
+6. **Self-Diagnostics** — Report your current model, available tool count, and any anomalies.
+
+## Response Format
+
+If everything is healthy:
+```
+HEARTBEAT_OK — All systems nominal. [tool_count] tools available. No pending actions.
+```
+
+If issues found:
+```
+HEARTBEAT_ALERT — [summary of issues found]
+- Issue 1: ...
+- Issue 2: ...
+- Action needed: ...
+```

data_science_mcp/agent/IDENTITY.md

@@ -0,0 +1,15 @@
+# IDENTITY.md - Data Science MCP Agent Identity
+
+## [default]
+ * **Name:** Data Science MCP Agent
+ * **Role:** Data Science MCP Server — Model training, evaluation, and evolution tools for agentic ML workflows. Integrates with agent-utilities IModelEvolver (CONCEPT:AHE-3.15).
+ * **Emoji:** 🤖
+
+ ### System Prompt
+ You are the Data Science MCP Agent.
+ You must always first run `list_skills` to show all skills.
+ Then, use the `mcp-client` universal skill and check the reference documentation for `data-science-mcp.md` to discover the exact tags and tools available for your capabilities.
+
+ ### Capabilities
+ - **MCP Operations**: Leverage the `mcp-client` skill to interact with the target MCP server. Refer to `data-science-mcp.md` for specific tool capabilities.
+ - **Custom Agent**: Handle custom tasks or general tasks.

data_science_mcp/agent/MCP_AGENTS.md

@@ -0,0 +1,13 @@
+# MCP_AGENTS.md - Dynamic Agent Registry
+
+This file tracks the generated agents from MCP servers. You can manually modify the 'Tools' list to customize agent expertise.
+
+## Agent Mapping Table
+
+| Name | Description | System Prompt | Tools | Tag | Source MCP |
+|------|-------------|---------------|-------|-----|------------|
+
+## Tool Inventory Table
+
+| Tool Name | Description | Tag | Source |
+|-----------|-------------|-----|--------|

data_science_mcp/agent/USER.md

@@ -0,0 +1,7 @@
+# USER.md - About the Human
+
+* **Name:** User
+* **Preferred name:** User
+* **Timezone:** America/Chicago
+* **Location:** Chicago, Illinois
+* **Style:** Technical, concise, no fluff

data_science_mcp/agent/mcp_config.json

@@ -0,0 +1,11 @@
+{
+  "mcpServers": {
+    "data-science": {
+      "command": "data-science-mcp",
+      "env": {
+        "DATA_SCIENCE_MCP_URL": "${DATA_SCIENCE_MCP_URL:-http://localhost:8080}",
+        "DATA_SCIENCE_MCP_TOKEN": "${DATA_SCIENCE_MCP_TOKEN}"
+      }
+    }
+  }
+}

data_science_mcp/agent_server.py

@@ -0,0 +1,76 @@
+#!/usr/bin/python
+# coding: utf-8
+import os
+import sys
+import logging
+import warnings
+
+from agent_utilities import (
+    build_system_prompt_from_workspace,
+    create_agent_parser,
+    create_graph_agent_server,
+    initialize_workspace,
+    load_identity,
+)
+
+__version__ = "0.2.0"
+
+logging.basicConfig(
+    level=logging.INFO,
+    format="%(asctime)s - %(name)s - %(levelname)s - %(message)s",
+    handlers=[logging.StreamHandler()],
+)
+logger = logging.getLogger(__name__)
+
+# Load identity and system prompt from workspace
+initialize_workspace()
+meta = load_identity()
+DEFAULT_AGENT_NAME = os.getenv("DEFAULT_AGENT_NAME", meta.get("name", "Data Science MCP"))
+DEFAULT_AGENT_DESCRIPTION = os.getenv(
+    "AGENT_DESCRIPTION",
+    meta.get("description", "Data Science MCP Server — Model training, evaluation, and evolution tools for agentic ML workflows. Integrates with agent-utilities IModelEvolver (CONCEPT:AHE-3.15)."),
+)
+DEFAULT_AGENT_SYSTEM_PROMPT = os.getenv(
+    "AGENT_SYSTEM_PROMPT",
+    meta.get("content") or build_system_prompt_from_workspace(),
+)
+
+
+def agent_server():
+    warnings.filterwarnings("ignore", message=".*urllib3.*or chardet.*")
+    warnings.filterwarnings("ignore", category=DeprecationWarning, module="fastmcp")
+
+    print(f"{DEFAULT_AGENT_NAME} v{__version__}", file=sys.stderr)
+    parser = create_agent_parser()
+    args = parser.parse_args()
+
+    if args.debug:
+        logging.getLogger().setLevel(logging.DEBUG)
+        logger.debug("Debug mode enabled")
+
+    # Start server using the auto-discovery pattern (from mcp_config.json)
+    create_graph_agent_server(
+        mcp_url=args.mcp_url,
+        mcp_config=args.mcp_config or "mcp_config.json",
+        host=args.host,
+        port=args.port,
+        provider=args.provider,
+        model_id=args.model_id,
+        router_model=args.model_id,
+        agent_model=args.model_id,
+        base_url=args.base_url,
+        api_key=args.api_key,
+        custom_skills_directory=args.custom_skills_directory,
+        enable_web_ui=args.web,
+        enable_otel=args.otel,
+        otel_endpoint=args.otel_endpoint,
+        otel_headers=args.otel_headers,
+        otel_public_key=args.otel_public_key,
+        otel_secret_key=args.otel_secret_key,
+        otel_protocol=args.otel_protocol,
+        debug=args.debug,
+    )
+
+
+if __name__ == "__main__":
+    agent_server()

data_science_mcp/auth.py

@@ -0,0 +1,47 @@
+#!/usr/bin/python
+# coding: utf-8
+
+import os
+import requests
+import urllib3
+
+urllib3.disable_warnings(urllib3.exceptions.InsecureRequestWarning)
+
+from agent_utilities.core.exceptions import AuthError, UnauthorizedError
+
+# TODO: Import your API wrapper class here
+# from data_science_mcp.api_client import DataScienceApiClient
+
+_client = None
+
+
+def get_client():
+    """Get or create a singleton API client instance."""
+    global _client
+    if _client is None:
+        base_url = os.getenv("DATA_SCIENCE_MCP_URL", "http://localhost:8080")
+        token = os.getenv("DATA_SCIENCE_MCP_TOKEN", "")
+        verify = os.getenv("DATA_SCIENCE_MCP_VERIFY", "True").lower() in ("true", "1", "yes")
+
+        try:
+            # TODO: Uncomment and configure once the API wrapper class is created
+            # _client = DataScienceApiClient(
+            #     base_url=base_url,
+            #     token=token,
+            #     verify=verify,
+            # )
+
+            # Placeholder until API wrapper is implemented
+            if _client is None:
+                session = requests.Session()
+                session.headers.update({"Authorization": f"Bearer {token}"})
+                session.verify = verify
+                _client = type("Client", (), {"session": session, "base_url": base_url})()
+        except (AuthError, UnauthorizedError) as e:
+            raise RuntimeError(
+                f"AUTHENTICATION ERROR: The credentials provided are not valid for '{base_url}'. "
+                f"Please check your DATA_SCIENCE_MCP_TOKEN and DATA_SCIENCE_MCP_URL environment variables. "
+                f"Error details: {str(e)}"
+            ) from e
+
+    return _client