Advanced · Systems

Adaptive Agent Context

Visual diagram · (in preparation) · Math · (in preparation) · Worked example · 3 difficulty levels.

TL;DR. Context that reshapes itself at runtime based on the task phase, difficulty, and what the agent has already learned.

Technical Definition

Context that reshapes itself at runtime based on the task phase, difficulty, and what the agent has already learned.

How it works

Adaptive context replaces a fixed prompt template with a policy that responds to signals: expand retrieval when the agent expresses uncertainty, load a debugging toolset once an error appears, drop planning scaffolding during mechanical execution, and increase the summarization rate as the window fills. The adaptation policy can be heuristic, learned, or delegated to the model itself through a context-management tool. The pay-off is spending tokens where they change the outcome instead of uniformly.

Related Concepts

• Agentic Context — The full set of information an agent has available when it decides its next action — goals, state, tools, memory, and retrieved data.
• Agentic Context Strategies — The repertoire of techniques for filling an agent's window: retrieval, summarization, scratchpads, sub-agents, and offloading to files.
• Context Management — The engineering practice of controlling what goes into a model's context window, in what order, and at what cost.
• Context-Aware Tool Selection — Exposing and choosing tools dynamically based on the current task phase rather than presenting every tool at every step.