agentsclimarketplace

Explore area

Skill huzaifa525/claude-code-optimizer/templates/.claude/skills/explore-area

Optimize Claude Code token usage — 22 skills, 8 hooks, 6 rules. One npm install. Zero manual steps. Cut costs by 67%.

Install
npx -y skills add huzaifa525/claude-code-optimizer --skill explore-area

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 9 stars9 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Use when you need to understand a module, feature, or directory before making changes.

SKILL.md

2.8 KB, as published. Nobody here has run it

Explore $ARGUMENTS using progressive disclosure to minimize token cost.

Progressive Disclosure (3 Layers)

Layer 1: File Index (cheapest — start here)

Get the file list WITHOUT reading contents:

# List all files in the target area
find [target] -type f -not -path '*/node_modules/*' -not -path '*/.git/*' | head -50

Or use Glob to find relevant files. From this list alone, identify:

  • Entry points (index., main., app.*)
  • Test files (.test., .spec.)
  • Config files
  • File count and general structure

Decision point: Do you now have enough context? If yes, skip to Return Format. If no, proceed to Layer 2.

Layer 2: Entry Points & Signatures (moderate cost)

Read ONLY entry points and key files. Use targeted reads:

  • Read index/main files to understand exports and module structure
  • Use Grep to find function signatures, class definitions, type exports
  • Map imports/exports to understand dependency graph
# Find exports
grep -rn "export" [target]/index.* 2>/dev/null
# Find class/function definitions
grep -rn "^export \(class\|function\|const\|interface\)" [target]/ 2>/dev/null

Decision point: Do you understand the module's API and structure? If yes, skip to Return Format. If no, proceed to Layer 3.

Layer 3: Deep Dive (expensive — only when needed)

Read full file contents, but only for files that are:

  • Directly relevant to the user's upcoming changes
  • Complex enough that signatures alone don't explain behavior
  • Containing patterns the user needs to follow

Use offset and limit for large files — read the relevant section, not the whole file.

Return Format

## Area: [name]

### Structure ([X] files)
- [file] → [purpose, 5 words max]

### Key Files (read these before editing)
- [file] → [what it does and why it matters]

### Dependencies
- Imports from: [external modules used]
- Exported to: [who consumes this module]

### Patterns to Follow
- [naming, structure, style patterns observed]
- [reference file] → follow this as template for new code

### Tests
- [test approach, test file locations, how to run]

### Gotchas
- [anything unusual, legacy code, known quirks]
- [things that look wrong but are intentional]

Rules

  • Start at Layer 1. Only go deeper when needed.
  • Never read ALL files in a directory — that defeats the purpose.
  • Focus on what someone needs to know BEFORE making changes.
  • Be concise. The user will read the files themselves when editing.
<!-- Skill by Huzefa Nalkheda Wala | github.com/huzaifa525 | claude-code-optimizer -->

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.