Module security 

Security validation utilities — SSRF prevention, path traversal detection, and tool description sanitization.

These utilities provide defense-in-depth against common input-based attacks targeting agentic AI systems. They map to OWASP LLM05 (Improper Output Handling / SSRF) and LLM01 (Prompt Injection).

Constants

MAX_DESCRIPTION_LEN

Maximum allowed length for tool descriptions.

MAX_TOOL_NAME_LEN

Maximum allowed length for tool names.

Functions

is_description_safe

Check whether a tool description contains prompt injection patterns.

is_env_path_safe

Validate a path from an env var for safety.

is_path_safe

Check whether a file path is safe from path traversal attacks.

is_path_within_base

Verify that a resolved path stays within the given base directory.

is_private_ip

Check if an IP address is in a private/reserved range.

is_tool_name_valid

Validate a tool name — must be alphanumeric with dots, underscores, hyphens.

is_url_safe

Check whether a URL is safe to fetch (SSRF prevention).

parse_clamped_f32

Parse an f32 from an env var string, clamping to a valid range.

parse_clamped_usize

Parse a usize from an env var string, clamping to a valid range.

sanitize_description

Sanitize a tool description by removing injection patterns.

sanitize_path

Sanitize a file path by removing dangerous sequences.