Ollama now supports subagents and web search directly within Claude Code, without requiring MCP servers or API keys. Users can launch Claude Code with any Ollama cloud model (e.g. minimax-m2.5:cloud, glm-5:cloud, kimi-k2.5:cloud) via a single CLI command. Subagents let tasks like file search, code exploration, and research run in parallel, each with its own context, keeping longer coding sessions from getting cluttered. Some models trigger subagents automatically, while others need explicit prompting. Web search is now built into Ollama's Anthropic compatibility layer, so models can fetch current information automatically, and subagents can combine parallel research with web search for tasks like auditing dependencies or comparing competitor pricing.
Questions this post answers
How do I use subagents with Claude Code through Ollama?
Launch Claude Code with an Ollama cloud model using the command ollama launch claude --model minimax-m2.5:cloud, then prompt the model to spawn or create subagents for tasks like exploring code, auditing security, or mapping database queries. Models such as minimax-m2.5, glm-5, and kimi-k2.5 can trigger subagents automatically, or you can force it by explicitly asking the model to use subagents. Each subagent runs in its own context so parallel tasks don't clutter the main session. Developers wiring up agentic coding workflows can track new subagent capabilities on daily.dev.
Do I need an API key or MCP server to use web search in Claude Code with Ollama?
No, web search is now built directly into Ollama's Anthropic compatibility layer, so no MCP servers or API keys are required. When a model needs current information, Ollama handles the search and returns results automatically without additional configuration, and subagents can use this web search to research topics in parallel. Anyone evaluating AI coding tool setups can follow integration changes like this on daily.dev.