broken-access-control-scanner 0.0.0

AI command line tool to detect broken access control vulnerabilities in source code

broken_access_control_scanner

broken_access_control_scanner is an AI command line tool to detect broken access control vulnerabilities in source code using Anthropic's Claude AI.

Installation

Releases are made available on PyPi. The recommended installation method is via pip:

pip install broken-access-control-scanner

Usage

python -m broken_access_control_scanner <source_file> --data-model "<data_model_description>"

Requires ANTHROPIC_API_KEY environment variable to be set.

Arguments

• source_file: Path to a source code file containing endpoints
• --data-model, -d: Description of the data model and context for the endpoints (required)
• --model, -m: Anthropic model to use (default: claude-sonnet-4-20250514)

Example

python -m broken_access_control_scanner api.py \
    --data-model "REST API with User and Document models. Users should only access their own profiles."

Output Example

┏━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
┃ Endpoint                  ┃  Severity  ┃ Description                      ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┩
│ /api/users/{id}/profile   │    NONE    │ Proper authorization check       │
│ /api/documents/{id}       │  CRITICAL  │ No authentication or auth check  │
└───────────────────────────┴────────────┴──────────────────────────────────┘

Severity Levels

• NONE: No access control issues found
• LOW: Minor issues, unlikely to be exploitable
• MEDIUM: Access control weakness that could be exploited under certain conditions
• HIGH: Clear access control vulnerability that can likely be exploited
• CRITICAL: Severe access control vulnerability with high impact, easily exploitable