context-architect

What problem does it solve?

Context-architect reduces token bloat and disorganized workspace docs by designing a clear, minimal document layer and system prompt strategy for AI agents and app builds. It helps teams decide what must load every session versus what should be on-demand, preventing duplication and costly always-loaded references.

Core Features & Use Cases

• Sequential interview: Conducts an adaptive, one-question-at-a-time interview to gather purpose, platform, existing docs, user profile, tasks, autonomy boundaries, and model strategy.
• Document map & drafts: Produces a prioritized file map with purposes, load rules, and estimated line counts, then writes full draft files one at a time.
• Token budget & audits: Estimates always-load token cost, recommends model routing, and flags anti-patterns such as mega-docs or duplication.
• Use case: Re-architect an existing assistant workspace to cut always-loaded context from 1,400 lines to a lean <400-line core and generate guardrails and routing docs for deployment.

Quick Start

Ask the skill to "Design my agent's context" by running a sequential interview to collect purpose, platform, existing docs, user profile, core tasks, autonomy boundaries, and model strategy, then return a document map, estimated token budget, anti-patterns found, and draft the core files.