Skip to main content
Glama
run_thinkdeep_web.py1.49 kB
import os, sys, asyncio, json # Ensure project root is on sys.path when running from scripts/ ROOT = os.path.abspath(os.path.join(os.path.dirname(__file__), os.pardir)) if ROOT not in sys.path: sys.path.insert(0, ROOT) from server import handle_call_tool # Configure environment for web mode and auto-continue (small) os.environ.setdefault('CLAUDE_DEFAULTS_USE_WEBSEARCH', 'true') os.environ.setdefault('EX_AUTOCONTINUE_WORKFLOWS', 'true') os.environ.setdefault('EX_AUTOCONTINUE_ONLY_THINKDEEP', 'true') os.environ.setdefault('EX_AUTOCONTINUE_MAX_STEPS', '2') async def main(): args = { 'step': ( 'Use provider-native web browsing to find current context windows and model aliases for: ' 'Moonshot/Kimi and ZhipuAI GLM. Focus on >=128k, 200k, 256k context and API enablement steps. ' 'Output: short bullet summary + exact model names/aliases/provider, and note any requirements to enable long context.' ), 'step_number': 1, 'total_steps': 2, 'next_step_required': True, 'findings': 'Plan web queries.', 'model': 'auto', 'use_websearch': True, 'thinking_mode': 'medium' } res = await handle_call_tool('thinkdeep', args) # Print the last text content block for i, item in enumerate(res): text = getattr(item, 'text', None) if text: print(f"\n=== BLOCK {i} ===\n{text[:4000]}\n") if __name__ == '__main__': asyncio.run(main())

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/Zazzles2908/EX_AI-mcp-server'

If you have feedback or need assistance with the MCP directory API, please join our Discord server