🍵 Ice Tea Security Scanner

AI-Powered Static Application Security Testing (SAST) — written in Go

---

Ice Tea is an advanced SAST tool that combines lightning-fast AST pattern matching (via Tree-sitter) with AI-powered deep reasoning (via OpenAI) to find security vulnerabilities in your source code — with dramatically fewer false positives than traditional scanners.

$ ice-tea scan ./src --format console --severity high

[CRITICAL] [CWE-89] SQL Injection — src/handlers/user.go:47
  Rule: GO-SQLI-001 | Confidence: high
  Code: query := "SELECT * FROM users WHERE id = " + userID

[CRITICAL] [CWE-798] Hardcoded JWT Secret — src/config/config.go:12
  Rule: AUTH-SEC-003 | Confidence: high
  Code: const jwtSecret = "supersecret123"

[HIGH]     [CWE-295] InsecureSkipVerify TLS — src/client/http.go:31
  Rule: CRYPTO-TLS-001 | Confidence: high
  Code: TLSClientConfig: &tls.Config{InsecureSkipVerify: true}

Summary: 3 findings (2 critical, 1 high) in 0.4s

---

Features

• 3-Engine Architecture:

  1. Pattern Matching Engine — AST + Regex rules via Tree-sitter and Go's native go/ast
  2. Taint Tracker — Traces data flow from user-controlled sources to dangerous sinks
  3. LLM Reasoning Engine — Optionally consults OpenAI to verify each finding and eliminate false positives

• 10 Languages: Go, JavaScript, TypeScript, Python, Java, PHP, Ruby, Rust, C, C++

• 82 Built-in SKILLs / 456+ Detection Rules across 12 security domains:

  • Authentication, Injection, Web/API, Cryptography, Filesystem, Infrastructure, Logging, Memory Safety, Cloud, Android, Network

• 4 Output Formats: Console, JSON, SARIF 2.1.0, GitLab SAST JSON

• MCP Server: Model Context Protocol integration for Claude / ChatGPT agentic workflows

• CI/CD Ready: Native SARIF output, configurable exit codes, .ice-tea.yaml config file

• Extensible SKILL System: Add custom detection rules in Markdown + YAML — no Go code required

---

Installation

Requirements: Go 1.21+, GCC or Clang (for Tree-sitter CGO bindings)

git clone https://github.com/zakirkun/ice-tea.git
cd ice-tea
make build

# Verify
./bin/ice-tea version

Windows (PowerShell):

# Install GCC via MSYS2: https://www.msys2.org/
git clone https://github.com/zakirkun/ice-tea.git
cd ice-tea
go build -o bin/ice-tea.exe ./cmd/ice-tea
.\bin\ice-tea.exe version

---

Quick Start

1. Scan a Project

# Scan with default settings (medium+ severity)
./bin/ice-tea scan ./your-project

# Show all findings including low severity
./bin/ice-tea scan ./your-project --severity low

# Only critical findings to keep CI fast
./bin/ice-tea scan ./your-project --severity critical

2. SARIF Output for GitHub

./bin/ice-tea scan ./src --format sarif --output results.sarif

# .github/workflows/security.yml
- name: Run Ice Tea
  run: ./bin/ice-tea scan ./src --format sarif --output results.sarif
  continue-on-error: true

- name: Upload SARIF
  uses: github/codeql-action/upload-sarif@v3
  with:
    sarif_file: results.sarif

3. AI False-Positive Filtering

export OPENAI_API_KEY="sk-proj-..."
./bin/ice-tea scan ./src --enable-llm --severity medium

4. Test Against Vulnerable Example Projects

# All 4 example apps (Go, PHP, Python, Node.js)
./bin/ice-tea scan ./examples --severity low

# Each individually
./bin/ice-tea scan ./examples/vulnerable-go
./bin/ice-tea scan ./examples/vulnerable-php --language php
./bin/ice-tea scan ./examples/vulnerable-python --language python
./bin/ice-tea scan ./examples/vulnerable-nodejs --language javascript

---

SKILL Coverage

Ice Tea ships with 82 SKILL definitions (456+ rules) covering:

Category	Skills	Key Detections
auth/	10	Hardcoded secrets, JWT weaknesses, session fixation, insecure cookies, OAuth flaws, MFA bypass
web/	16	XSS, SSRF, CSRF, CORS misconfiguration, clickjacking, prototype pollution, ReDoS, WebSocket
injection/	9	SQL, Command, XXE, Template (SSTI), LDAP, XPath, NoSQL, Header/CRLF, Log injection
crypto/	8	Weak hash (MD5/SHA1), weak cipher (DES/RC4/ECB), insecure TLS, hardcoded IV, insecure RSA
api/	7	BOLA/IDOR, broken function-level auth, missing rate limiting, GraphQL security, mass assignment
fs/	5	Path traversal, unsafe file upload, Zip Slip, unsafe temp files
infra/	5	Debug mode in production, default credentials, insecure Docker/Kubernetes, hardcoded IPs
logging/	3	Sensitive data in logs, insufficient security logging, verbose error responses
memory/	5	Buffer overflow, use-after-free, null pointer dereference, integer overflow, format string
cloud/	5	AWS/GCP/Azure misconfigurations, secrets in env files, IaC security (Terraform)
android/	5	Insecure data storage, exported components, WebView JS bridge, intent redirection
network/	4	Cleartext HTTP, insecure sockets, DNS rebinding, extended SSRF (cloud metadata)

---

Creating Custom SKILLs

Adding a new detection rule requires only two files — no Go code needed:

skills/
└── myteam/
    └── my-custom-vuln/
        ├── SKILL.md      ← Metadata + remediation guidance
        └── patterns.yaml ← Detection rules (regex or AST)

SKILL.md (frontmatter + markdown):

---
name: My Custom Vulnerability
version: 1.0.0
description: Detects unsafe use of dangerous_function()
tags: [custom, injection]
languages: [python]
severity: high
confidence: high
cwe: [CWE-78]
owasp: [A03:2025]
---
# My Custom Vulnerability
## Overview
...
## Remediation
...

patterns.yaml:

rules:
  - id: CUSTOM-001
    name: dangerous-function-call
    description: "dangerous_function() called with user input"
    severity: high
    confidence: high
    languages: [python]
    patterns:
      - regex: "dangerous_function\\s*\\(.*request\\."
      - ast_node_type: "call_expression"
        function: "dangerous_function"

Then scan with your custom skills:

./bin/ice-tea scan ./src --skills-dir ./my-skills

📖 Read the full guide: docs/08-how-to-create-skills.md

---

Configuration File

Place .ice-tea.yaml in your project root:

output:
  format: "sarif"
  file: "results.sarif"

scan:
  severity: "medium"
  confidence: "medium"
  concurrency: 8

exclude:
  dirs: ["vendor", "node_modules", "testdata", "dist"]
  files: ["*.min.js", "*_test.go", "*.pb.go"]

llm:
  enabled: false
  provider: "openai"
  model: "gpt-4o"

---

Exit Codes

Code	Meaning
0	No findings above threshold
1	Findings found above threshold
2	Fatal error (bad args, config issue, etc.)

---

Documentation

Document	Description
Architecture Overview	System design and engine pipeline
AI Agent Skills	How the SKILL system works
AST Parsing	Tree-sitter and Go AST internals
Vulnerability Detection	Pattern matching and taint tracking
CI/CD Integration	GitHub Actions, GitLab CI, Jenkins
MCP Integration	Claude / ChatGPT agent workflows
How to Create SKILLs	Step-by-step guide for custom rules
Usage Guide	Complete CLI reference and examples

---

Contributing

Contributions are welcome — especially new SKILL definitions! See CONTRIBUTING.md for guidelines on:

• Adding new vulnerability patterns
• Reporting false positives
• Improving parser coverage
• Submitting pull requests

---

License

This project is licensed under the MIT License — see the LICENSE file for details.

Copyright (c) 2026 Ice Tea Scanner Contributors