@nlci/cli

Command-line interface for NLCI - Neural-LSH Code Intelligence.

Installation

npm install -g @nlci/cli
# or
pnpm add -g @nlci/cli

Quick Start

# Initialize NLCI in your project
nlci init

# Scan your codebase to build the index
nlci scan ./src

# Query for similar code
nlci query --file ./src/my-function.ts

# Generate a clone detection report
nlci report

Commands

nlci init

Initialize NLCI configuration in the current directory.

nlci init [options]

Options:
  -f, --force       Overwrite existing configuration
  --typescript      Use TypeScript configuration file
  --json            Use JSON configuration file

nlci scan

Scan a directory and build the code similarity index.

nlci scan [path] [options]

Arguments:
  path                    Path to scan (defaults to current directory)

Options:
  -o, --output <path>     Output path for index file (default: ".nlci-index")
  -c, --config <path>     Path to configuration file
  -i, --include <patterns...>   Glob patterns to include
  -e, --exclude <patterns...>   Glob patterns to exclude
  --min-tokens <n>        Minimum tokens per code block (default: 10)
  --max-tokens <n>        Maximum tokens per code block (default: 10000)
  -l, --languages <langs...>    Languages to include
  -v, --verbose           Show detailed output
  -f, --force             Force rebuild of existing index

nlci query

Query for similar code blocks.

nlci query [options]

Options:
  -x, --index <path>      Path to index file (default: ".nlci-index")
  -f, --file <path>       Query using code from a file
  -c, --code <code>       Query using inline code snippet
  -t, --threshold <value> Minimum similarity threshold (default: 0.85)
  -n, --limit <n>         Maximum number of results (default: 10)
  --type <type>           Filter by clone type (1, 2, 3, 4)
  --format <format>       Output format: table, json, compact (default: table)
  -v, --verbose           Show detailed output

nlci report

Generate a code clone report.

nlci report [options]

Options:
  -x, --index <path>      Path to index file (default: ".nlci-index")
  -o, --output <path>     Output file path
  -f, --format <format>   Report format: console, json, html, markdown
  -t, --threshold <value> Minimum similarity threshold (default: 0.85)

nlci serve

Start a local API server for code similarity queries.

nlci serve [options]

Options:
  -x, --index <path>      Path to index file (default: ".nlci-index")
  -p, --port <port>       Port to listen on (default: 3000)
  -h, --host <host>       Host to bind to (default: localhost)

API Endpoints:

• GET /health - Health check
• GET /stats - Index statistics
• POST /query - Query for similar code
• POST /clones - Get all detected clones

nlci stats

Display index statistics.

nlci stats [options]

Options:
  -x, --index <path>      Path to index file (default: ".nlci-index")
  --json                  Output as JSON

Configuration

NLCI looks for configuration in these locations (in order):

1. nlci.config.js / nlci.config.ts
2. nlci.config.json
3. .nlcirc / .nlcirc.json
4. package.json (nlci field)

Example Configuration

// nlci.config.js
module.exports = {
  lsh: {
    numTables: 20, // Number of hash tables (L)
    numBits: 12, // Bits per hash (K)
    dimension: 384, // Embedding dimension
  },
  embedding: {
    model: 'sentence-transformers/all-MiniLM-L6-v2',
    batchSize: 32,
    maxLength: 512,
  },
  parser: {
    minBlockSize: 10,
    maxBlockSize: 10000,
    includePatterns: ['**/*.ts', '**/*.js'],
    excludePatterns: ['**/node_modules/**'],
  },
};

Clone Types

| Type | Similarity | Description | | ------ | ---------- | -------------------------------------------------- | | Type-1 | ≥99% | Exact clones (whitespace/comment differences only) | | Type-2 | 95-99% | Parameterized clones (identifier renaming) | | Type-3 | 85-95% | Near-miss clones (statement modifications) | | Type-4 | 70-85% | Semantic clones (different syntax, same logic) |

Performance

• Indexing: O(n) - Linear time to scan codebase
• Query: O(1) - Constant time lookups using LSH
• Memory: Efficient bucket-based storage

License

AGPL-3.0-or-later