Blog
Mert Özen's blog posts on software development, backend, frontend, and DevOps.
Subagents in Claude Code: Splitting Context, Specializing and Running in Parallel
In Claude Code a subagent is a separate Claude instance that the main session hands a side task to and that works in its own context window: it doesn't see the conversation history, does see CLAUDE.md, and returns only a summary. The .claude/agents/ file format, the tools allowlist, the model and memory fields, the background and forks, worktree isolation, and ef-query-reviewer, angular-a11y-checker and test-runner for Bookstore.
Plan Mode and the Daily Workflow: Explore, Plan, Implement, Verify and Manage Context
The safe way to have Claude Code build a large feature is to get a plan with a file list in plan mode, fix it with Ctrl+G, approve it and commit after every verified layer; checkpoints don't roll back files changed by Bash or the database. Plan approval options, a completion condition with /goal, compaction and Compact Instructions, /clear, /resume, /branch and /rewind.
The LLM Context Window: What the Model Remembers and What It Silently Cuts
LLM APIs are stateless; you send the chat history with every request and every turn grows the bill. Even though Gemma 3 supports 128K tokens, Ollama allocates a 4096-token window by default and cuts a 4498-token prompt down to 2051 tokens without any error. We set num_ctx in C# and TypeScript and try managing history with trimming and summarizing, using real output.