How does Context Packer save tokens?

It extracts only the most important metadata—such as file maps and function signatures—instead of sending full file contents, significantly reducing the token count per prompt.

Can it see my recent git history?

Yes, Context Packer includes the subjects of your last three relevant commits to help Claude understand the most recent changes and project direction.

Does it still provide actual code samples?

Yes, but it is strictly limited to one short code window only if it is absolutely critical to the task, preferring pointers to file paths and line numbers instead.

Is Context Packer compatible with large repositories?

Absolutely. In fact, it is specifically designed for large repos where sending the entire codebase would be impossible or prohibitively expensive.

When is the best time to trigger this skill?

Use it during 'UserPromptSubmit' enrichment or right before calling a sub-agent to ensure the AI has the right architectural context without the noise.

Context Packer

Name: Context Packer
Author: MacroMan5

byMacroMan5

•

生产力与工作流

Optimizes AI token usage by generating compact, high-signal context briefs of files, symbols, and recent repository changes.

关于

Context Packer is a specialized efficiency tool for Claude Code designed to prevent context bloat and minimize token consumption. It replaces the habit of pasting large code blocks with a dense, 10–20 line architectural brief that includes critical file maps, key symbols, and recent commit history. By providing Claude with a high-level overview instead of raw data, the skill improves response speed, reduces costs, and helps the AI maintain a clear understanding of the project structure without hitting context window limits.

主要功能

Maps key file paths and directory structures
2 GitHub stars
Identifies primary symbols, functions, and classes
Enforces strict guardrails against large code injections
Generates dense 10–20 line project summaries
Summarizes the last three relevant git commits

使用场景

Summarizing repository state before implementing complex features like auth middleware
Onboarding sub-agents to a specific module with minimal token overhead
Providing high-level context for code reviews without sending entire source files

关于

主要功能

Maps key file paths and directory structures
2 GitHub stars
Identifies primary symbols, functions, and classes
Enforces strict guardrails against large code injections
Generates dense 10–20 line project summaries
Summarizes the last three relevant git commits

使用场景

Summarizing repository state before implementing complex features like auth middleware
Onboarding sub-agents to a specific module with minimal token overhead
Providing high-level context for code reviews without sending entire source files