token-architect

Reduce token usage and maintain context during multi-file coding tasks.

Updated Apr 19, 2026
One-click install
npx skills add https://github.com/cjrfoto/claude-config --skill token-architect
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: token-architect
Source: https://github.com/cjrfoto/claude-config/tree/main/skills/token-architect
Command: npx skills add https://github.com/cjrfoto/claude-config --skill token-architect

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps developers manage and minimize token usage, maintain context hygiene, and avoid hitting session limits during large multi-file coding tasks.

Core Features & Use Cases

  • Context hygiene: maintain concise, relevant context across files to prevent token bloat.
  • Plan-before-code discipline: generate a plan before making changes to reduce unnecessary edits.
  • Deterministic task framing: provide clear, bounded instructions to keep AI actions aligned with project scope.

Quick Start

Describe the files you will modify and confirm the plan before starting heavy multi-file coding tasks.

Frequently Asked Questions about token-architect

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I optimize token usage during heavy multi-file coding tasks?

To optimize token usage during multi-file coding, enforce deterministic planning and context hygiene checks before modifying files. This prevents token bloat by maintaining concise, relevant context and generating a plan to reduce unnecessary edits.

What is context hygiene in AI-assisted coding workflows?

Context hygiene in AI-assisted coding is the practice of maintaining concise, relevant context across files to prevent token bloat. It ensures that AI actions remain aligned with project scope by providing clear, bounded instructions.

Why does my AI coding session hit token limits when working with large codebases?

AI coding sessions hit token limits due to context bloat from unmanaged multi-file edits. Implementing plan-before-code discipline and deterministic task framing reduces unnecessary edits and preserves token budgets during heavy coding.

Can I enforce deterministic task framing for sub-tasks in large codebase modifications?

Yes, you can enforce deterministic task framing by providing clear, bounded instructions for sub-tasks. This safe delegation keeps AI actions aligned with project scope and maintains context hygiene across multi-file modifications.

What's the best way to plan before coding to reduce unnecessary edits?

The best way to plan before coding is to describe the files you will modify and confirm the plan before starting heavy tasks. This plan-before-code discipline generates bounded instructions, reducing unnecessary edits and saving tokens.

Does deterministic planning work for safe delegation of sub-tasks in coding workflows?

Deterministic planning works for safe delegation by providing clear, bounded instructions for sub-tasks. This approach keeps AI actions aligned with the project scope and maintains context hygiene during multi-file coding workflows.