Specter Scan

Build or rebuild the knowledge graph for the current codebase.

When to Use

• •First time using Specter on a project
• •After major refactoring or adding many files
• •When specter tools return stale or missing data
• •Before asking the specter agent questions about the codebase

How to Run

Execute the specter scan command:

npx specter-mcp scan

Or with options:

# Skip git history analysis (faster)
npx specter-mcp scan --no-git

# Force rescan even if graph exists
npx specter-mcp scan --force

# Scan a specific directory
npx specter-mcp scan --dir ./packages/core

What It Does

1. •Parses all TypeScript/JavaScript files using AST analysis
2. •Extracts functions, classes, interfaces, types, and exports
3. •Maps import relationships between files
4. •Calculates cyclomatic complexity for each function
5. •Analyzes git history for modification patterns (unless --no-git)
6. •Saves the knowledge graph to .specter/graph.json

Output

After scanning, you'll see:

• •File count and total lines
• •Symbol count (functions, classes, etc.)
• •Relationship count (imports)
• •Complexity metrics
• •Language breakdown

The .specter/ directory will contain:

• •graph.json — The full knowledge graph
• •metadata.json — Quick-access scan statistics

Performance

• •~1000 files per minute on typical hardware
• •Git history adds ~30% scan time
• •Use --no-git for faster iteration during development

Notes

• •Only TypeScript and JavaScript files are analyzed
• •node_modules, dist, and build directories are excluded
• •Test files (*.test.ts, *.spec.ts) are excluded
• •The .specter/ directory is automatically added to .gitignore