Tagged context window · show all articles
A long agent loop resends its own history on every step, pays a fixed tool tax each time, and risks a retry that resends it twice. That arithmetic, not model quality, sets the real ceiling on how far a trajectory can run.
OpenAI, Anthropic, and Google each publish a token count for how much a model can hold. Independent long-context benchmarks measure a different quantity: how much of it the model actually uses correctly, and the two numbers are routinely far apart.
CLAUDE.md looks like a settings hierarchy — enterprise, user, project, local. It is not one. Claude Code concatenates every discovered file into a fixed token budget and reads the result as a prompt, not a policy.
Claude Code is not a chatbot with a code icon. It is an agentic harness: a loop that calls a model, executes real shell and filesystem operations behind a permission gate, and manages a shrinking context budget one turn at a time.
Claude's context window now reaches a million tokens, but independent benchmarks that vary length, position, and lexical overlap show recall degrading well before the window fills — evidence the model's own maker now calls context rot.
Retrieval-augmented generation is usually described as one step: ask, then answer. In production it is six separate stages, each with its own literature, its own failure mode, and its own knob to turn.