context-cartography

What problem does it solve?

It solves the problem of designing what an agent should include in its context window so the agent has the right information to complete the task without wasting tokens or getting distracted by irrelevant material.

Core Features & Use Cases

• Boundary-aware context design: Helps you decide when to redesign context, when to triage by cutting, and when not to use the skill for one-off or capability-focused prompts.
• Six-step workflow (greenfield): Guides task definition, source surveying, prioritization (P0–P3), sizing, structuring, and cutting into an explicit context plan.
• Triage-first workflow (context too large): Starts from CUT and works backward when removals reveal that important priorities were accidentally deprioritized.
• Context Manifest output for validation: Produces a versionable, testable manifest (including token budget, structure order, shadow context) designed to be validated with EDD/context-eval.

Quick Start

Use context-cartography to design a context manifest for your agent by asking, "Review pull requests for a Python microservices codebase and decide exactly what context sources to include and cut so the agent stays accurate and secure."