code-intelligence

Original：🇺🇸 English

Translated

8 scripts

Semantic code search, relationship mapping, and codebase intelligence. Provides deep code analysis including symbol search, dependency tracking, change impact analysis, and codebase Q&A.

4installs

Sourcewinsorllc/upgraded-carnival

Added on2026-04-16

NPX Install

npx skill4agent add winsorllc/upgraded-carnival code-intelligence

SKILL.md Content

View Translation Comparison →

Code Intelligence Skill

Advanced codebase analysis and search tools inspired by hybrid memory systems. This skill enables the agent to understand code relationships, search semantically, and analyze the impact of changes.

Setup

bash

cd /job/.pi/skills/code-intelligence
npm install

Tools

Index Codebase

Build a searchable index of the codebase:

bash

node index.js --scan /path/to/project

This creates:

Symbol table (functions, classes, exports)
Import/dependency graph
File metadata index

Semantic Search

Search for code by meaning, not just text:

bash

# Find code related to authentication
node search.js "authentication middleware"

# Find where database connections are established
node search.js "database connection setup"

# Search with filters
node search.js "user validation" --type function --ext js,ts

Relationship Analysis

Understand code dependencies:

bash

# What depends on this file?
node deps.js --dependents src/utils/api.js

# What does this file import?
node deps.js --imports src/utils/api.js

# Show full dependency chain
node deps.js --chain src/app.js

Impact Analysis

Predict what might break with changes:

bash

# Analyze impact of modifying a function
node impact.js src/utils/auth.js --function verifyToken

# Check impact of file changes
node impact.js src/config/database.js

Code Q&A

Ask questions about the codebase:

bash

node ask.js "Where is the JWT secret configured?"
node ask.js "How does error handling work in this project?"
node ask.js "What functions use the database connection?"

Architecture

The skill uses a layered approach:

Parser Layer: Tree-sitter for AST parsing
Index Layer: SQLite for fast lookups + JSON for graph data
Search Layer: Hybrid keyword + semantic matching
Analysis Layer: Graph algorithms for relationship mapping

Output Format

All tools output JSON for programmatic use or formatted text for readability:

bash

# JSON output (for piping to other tools)
node search.js "auth" --json

# Formatted text (default)
node search.js "auth"

When to Use

Understanding large or unfamiliar codebases
Finding all usages of a function or variable
Assessing risk before making changes
Discovering patterns across the codebase
Answering "where is X?" questions
Code review preparation
Refactoring planning

Requirements

Node.js ≥18
Tree-sitter parsers for supported languages (auto-installed)

Supported Languages

JavaScript/TypeScript
Python
Rust
Go
Java
Ruby
And more via tree-sitter

Data Storage

Index files are stored in

.code-intelligence/

:

```
index.db
```
- SQLite database
```
graph.json
```
- Dependency graph
```
cache/
```
- Parsed AST cache

Add

.code-intelligence/

to your

.gitignore

.

code-intelligence

NPX Install

Tags

SKILL.md Content

Code Intelligence Skill

Setup

Tools

Index Codebase

Semantic Search

Relationship Analysis

Impact Analysis

Code Q&A

Architecture

Output Format

When to Use

Requirements

Supported Languages

Data Storage