APILeak v1.0.0

APILeak v1.0.0 — Enterprise API Security Testing Tool

Enterprise-grade API security testing tool with full OWASP API Security Top 10 2023 coverage, JWT attack suite, cloud gateway auditing, and an MCP server that lets AI agents run scans via natural language.

---

📖 Documentation

Full docs site: https://moisestapia.github.io/apileaks/

Guide	Description
MCP Server Guide	Connect Claude / Kiro and run scans via AI
Docker Deployment	Docker & docker-compose for the MCP server
AI Providers	Anthropic, OpenAI, Bedrock configuration
MCP Tools Reference	JSON Schema for every MCP tool
OWASP Modules Guide	All 12 OWASP testing modules
JWT Attack Testing	8-attack JWT test suite
MCP Security	OWASP MCP Top 10 controls

---

🚀 Quick Start

# Install
pip install 'apileaks[mcp,ai]'

# Or from source
git clone https://github.com/MoisesTapia/apileaks.git
cd apileaks
pip install -e '.[mcp,ai]'

# Run a scan
apileaks scan --target https://api.example.com

# See all commands
apileaks --help

---

🤖 MCP Server — AI Agent Integration

APILeak v1.0.0 exposes every scan capability as an MCP tool. Connect Claude Desktop, Kiro, or any MCP-compatible AI agent and run full API security tests via natural language.

Connect Claude Desktop (stdio — 1 config line)

Edit ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

{
  "mcpServers": {
    "apileaks": {
      "command": "apileaks",
      "args": ["mcp-server", "--transport", "stdio"]
    }
  }
}

Restart Claude. The apileaks toolset appears automatically.

Connect Kiro

In .kiro/settings/mcp.json:

{
  "mcpServers": {
    "apileaks": {
      "command": "apileaks",
      "args": ["mcp-server", "--transport", "stdio"]
    }
  }
}

HTTP server (persistent, multi-client)

export APILEAKS_MCP_AUTH_TOKEN=$(python -c "import secrets; print(secrets.token_urlsafe(32))")
apileaks mcp-server --transport http --port 8080

Docker Compose

echo "APILEAKS_MCP_AUTH_TOKEN=change-me" > .env
docker compose --profile mcp up -d
# Server available at http://localhost:8080/sse

What the AI agent can do

Once connected, you can ask your AI agent:

"Run a full OWASP scan against https://staging.api.com using Bearer token eyJ..." "Test the /users/{id} endpoint for BOLA. Check IDs 1 through 100." "Audit the JWT implementation. Here's a sample token: eyJhbGci..."

The AI invokes the appropriate tools (apileak.scan, apileak.owasp.bola_testing, apileak.jwt.attack_test, etc.) and returns structured findings.

Security: Destructive tools require a ConfirmationToken — the AI cannot execute them without explicit human confirmation. Every invocation is logged to an append-only audit trail (OWASP MCP08).

---

📋 Command Overview

apileaks --help

Command	Description
apileaks req / apileaks request	Send a single HTTP request (coloured dump, proxy, secret scan)
apileaks scan	Orchestrated scan — discovery + all (or selected) OWASP modules
apileaks owasp <module>	Run a single OWASP module in isolation
apileaks dir	Directory/endpoint fuzzing with triage
apileaks par	Hidden parameter discovery
apileaks brute	Brute-force exposed API spec files
apileaks jwt <subcommand>	JWT attack & utility toolkit (13 subcommands)
apileaks cloud-audit	Cloud API gateway security audit (AWS / GCP / Azure)
apileaks mcp-server	Start the MCP server (stdio or HTTP/SSE)
apileaks mcp-audit	Audit a third-party MCP server for security issues
apileaks replay <report>	Re-issue requests from a prior scan report
apileaks wordlist	Manage Assetnote wordlists

---

✨ Features

OWASP API Security Top 10 (2023) — 12 modules

Module	CLI key	OWASP Category
BOLA / IDOR	bola	API1:2023
Broken Authentication	auth	API2:2023
Mass Assignment	property	API3:2023
Resource Consumption	resource	API4:2023
BFLA	function	API5:2023
Business Flows	business	API6:2023
SSRF	ssrf	API7:2023
Security Misconfiguration	security	API8:2023
Inventory Management	inventory	API9:2023
Unsafe Consumption	unsafe	API10:2023
HTTP Parameter Pollution	parameter_pollution	(beyond Top 10)
OpenAPI Contract Audit	spec_driven_testing	(beyond Top 10)

JWT Attack Suite (8 attacks)

apileaks jwt attack-test --target https://api.example.com/protected --token eyJ...

• None algorithm bypass
• Weak secret brute-force (HS256/384/512)
• RS256 → HS256 algorithm confusion
• JWK header injection
• kid SQL injection / path traversal
• Expired token replay
• sub claim manipulation

AI-Assisted Features (optional — apileaks[ai])

# LLM finding triage and prioritisation
apileaks scan --target URL --ai-triage

# Semantic deduplication of overlapping module findings
apileaks scan --target URL --ai-dedup

# Both
apileaks scan --target URL --ai-triage --ai-dedup

Supports Anthropic (Claude), OpenAI (GPT), and Amazon Bedrock. Set the provider in config/apileaks.yaml + the corresponding API key env var. Graceful degradation: if the provider is unavailable, the scan completes without the AI section.

Cloud API Gateway Audit

# AWS API Gateway
apileaks cloud-audit --provider aws \
  --target https://api-id.execute-api.us-east-1.amazonaws.com

# Google Cloud API Gateway
apileaks cloud-audit --provider gcp --target https://gateway.example.com

# Azure API Management
apileaks cloud-audit --provider azure \
  --target https://apim-name.azure-api.net

MCP Security Self-Audit

# Audit any MCP server (including yourself)
apileaks mcp-audit --target "apileaks mcp-server --transport stdio"
apileaks mcp-audit --target http://mcp.example.com --fail-on-findings

Checks: tool poisoning, SSRF via URL params, missing confirmation gates, credential leaks in responses.

Single HTTP Request (apileaks req)

Send a single request with a rich coloured dump, native proxy support, secret scanning, and session persistence:

# Basics
apileaks req GET  https://api.example.com/users
apileaks req POST https://api.example.com/users name=alice role=admin
apileaks req POST https://api.example.com/users --json '{"name":"alice","role":"admin"}'

# Through Burp / Caido / Hetty
apileaks req PATCH https://api.example.com/users/42 \
  --json '{"role":"admin"}' --proxy http://127.0.0.1:8080

# Secret scan on response
apileaks req GET https://api.example.com/config --scan-response

# Pipe response body
apileaks req GET https://api.example.com/token --quiet | jq .access_token

# Save for replay
apileaks req POST https://api.example.com/users name=alice \
  --save-session sessions/pentest.json
apileaks replay sessions/pentest.json --proxy http://127.0.0.1:8080

---

Full scan with CI gate

apileaks scan \
  --target https://api.example.com \
  --modules bola,auth,ssrf,property \
  --openapi https://api.example.com/openapi.json \
  --ci-mode \
  --fail-on high \
  --sarif \
  --output-file report.sarif

OWASP modules individually

# List all modules with descriptions
apileaks owasp

# BOLA — test horizontal privilege escalation
apileaks owasp bola --target https://api.example.com --auth-token "Bearer eyJ..."

# SSRF — probe URL-accepting parameters
apileaks owasp ssrf --target https://api.example.com

# Spec contract audit (OpenAPI)
apileaks owasp spec_driven_testing \
  --target https://api.example.com \
  --openapi https://api.example.com/openapi.json

Discovery

# Directory fuzzing
apileaks dir --target https://api.example.com --wordlist wordlists/endpoints.txt

# With Assetnote wordlists
apileaks wordlist fetch apiroutes-210328
apileaks dir --target https://api.example.com --wordlist assetnote:apiroutes-210328:20000

# Parameter fuzzing
apileaks par --target https://api.example.com/search

# API spec brute-force
apileaks brute --target https://api.example.com

Advanced scan options

# Framework detection + version fuzzing
apileaks scan --target URL --detect-framework --fuzz-versions

# WAF evasion
apileaks scan --target URL --user-agent-random --enable-waf-evasion

# Multi-target from file
apileaks scan --target-file hosts.txt --max-hosts 50

# Replay requests from a previous scan
apileaks replay reports/scan_2026-11-03.json

---

🐳 Docker

# Base image (scanner only)
docker build -t apileak:latest .
docker run --rm apileak:latest scan --target https://api.example.com

# MCP server image
docker build -f Dockerfile.mcp -t apileak-mcp:latest .

# docker-compose — full MCP stack
APILEAKS_MCP_AUTH_TOKEN=secret docker compose --profile mcp up -d

# Run the security self-audit
docker compose --profile mcp-audit up --abort-on-container-exit

See Docker Deployment Guide for TLS, reverse proxy, and CI/CD details.

---

📦 Installation

Requirements

• Python 3.11+
• pip / uv / pipx

Extras

Extra	Installs	Use for
apileaks[mcp]	MCP SDK, jsonschema, starlette, uvicorn	mcp-server, mcp-audit
apileaks[ai]	anthropic, openai	--ai-triage, --ai-dedup
apileaks[ai,aws]	+ boto3	Amazon Bedrock provider
apileaks[gcp]	google-cloud-api-gateway	GCP cloud audit
apileaks[azure]	azure-identity, azure-mgmt	Azure cloud audit
apileaks[dev]	pytest, hypothesis, ruff, mypy	Development

# Minimal (scanner only)
pip install apileaks

# MCP + AI
pip install 'apileaks[mcp,ai]'

# Everything
pip install 'apileaks[mcp,ai,aws,gcp,azure]'

---

🔒 Security

• All 10 OWASP MCP Top 10 controls implemented in the MCP server
• Append-only audit log for every tool invocation
• Destructive tools require explicit human ConfirmationToken
• Hash-verified MCP SDK supply chain (requirements-mcp.txt)
• CI self-audit: the MCP server audits itself before every release

See MCP Security Documentation and SECURITY.md.

---

🤝 Contributing

See CONTRIBUTING.md for development setup, coding standards (ruff, mypy, pytest), MCP SDK version bump policy, and PR workflow.

📄 License

MIT License — free to use, modify, and distribute.

---

Docs · Releases · Issues

Built for API security professionals.