ham

Scope CLAUDE.md context to active directories to reduce per-prompt token usage.

65|3|Updated Feb 23, 2026

Search Tags:#automation #context #optimization #dashboard #memory #token #ham

Ready to run

npx skills add https://github.com/kromahlusenii-ops/ham --skill ham

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: ham
Source: https://github.com/kromahlusenii-ops/ham/tree/main
Command: npx skills add https://github.com/kromahlusenii-ops/ham --skill ham

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

HAM reduces per-prompt token costs by scoping context to the active directory rather than loading a monolithic memory file.

Core Features & Use Cases

Scoped CLAUDE.md per directory to minimize token usage and improve performance
On-demand memory loading: root + target directory context for each task
Integrated analytics: token savings, dashboard, and baseline benchmarking

Quick Start

Run go ham in your repository to auto-generate the scoped CLAUDE.md memory structure.

Frequently Asked Questions about ham

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

FAQPage Schema

How do I reduce token usage in large codebases when loading context for AI agents?▼

Reduce per-prompt token usage by scoping CLAUDE.md context to the active directory. This approach loads only root plus target directory memory instead of a monolithic file, minimizing costs and improving agent performance across large codebases.

How does scoped memory loading work for multiple agents in a repository?▼

Scoped memory loading enforces per-directory CLAUDE.md creation, loading root and target directory context on demand. It applies across large codebases and multiple agents, ensuring each task receives only the necessary scoped context to answer efficiently.

What is the best way to generate scoped CLAUDE.md memory structures?▼

Generate scoped CLAUDE.md memory structures by running the command in your repository. This auto-generates the scoped memory structure, enforcing per-directory file creation to minimize token usage and improve context loading performance.

Can I track token savings and carbon efficiency for AI context optimization?▼

Track token savings and carbon efficiency using the integrated dashboard and baseline tooling. HAM provides analytics for token savings and baseline benchmarking, allowing you to monitor context optimization and carbon efficiency across codebase directories.

Does context scoping work with on-demand memory files across different directories?▼

Context scoping supports on-demand memory files across different directories. HAM enforces per-directory CLAUDE.md creation and loads root plus target directory context on demand, applying efficiently across large codebases and multiple agents.