Back to skill

Security audit

Axiom Url Canonicalizer

Security checks for vulnerabilities and agentic risk

Overview

This skill is a local URL-normalization utility and I did not find hidden network, persistence, credential, or destructive behavior.

Reasonable to install for local URL canonicalization. Be aware that the manifest is stale and documentation is mixed-language; do not treat the tool as a URL safety validator or redirect checker.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (6)

Scope Creep

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The manifest describes the package as a simple file inventory and does not disclose that the skill performs live HTTP requests. This mismatch can cause reviewers or operators to grant trust under the assumption of passive string processing, when the code actually has network side effects such as outbound requests, SSRF exposure, privacy leakage, or policy noncompliance.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Low
Category
Not specified by scanner
Confidence
79% confidence
Finding

Branding the skill as 'public, clean' and a URL canonicalizer may understate that canonicalization includes active fetching rather than purely local normalization. In security-sensitive environments, this can mislead users about the skill's behavior and increase the chance it is used with untrusted input in contexts where outbound requests are unsafe.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill description and usage guidance are written in French (for example, 'Problème résolu', 'Cas d'usage', and 'Limitations') with no indication that users may choose another language. The policy scope includes natural-language locale violations in markdown, and this file does not document a justified region-specific constraint or user opt-in.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
95% confidence
Finding

This code file contains user-facing natural language such as "LIMITATIONS CONNUES", "NORMALISE LES URLs POUR SEO ET CACHE", and CLI usage/help text in French. The policy allows locale constraints only when users are given a choice or when the constraint is clearly justified, neither of which is present here.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
88% confidence
Finding

The module-level documentation states there is no support for 'javascript: / mailto: / data: schemes'. However, the code explicitly includes 'mailto' in LOWERCASE_SCHEMES, and the canonicalization logic will parse and rebuild such URLs rather than rejecting them. This is an intent/documentation contradiction rather than a mere omission.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
91% confidence
Finding

The docstring contains user-facing operational text such as limitations and CLI usage in French. The policy requires avoiding forced language or locale constraints unless the skill offers a choice or clearly justifies the locale, which is not present here.

Content

No source excerpt is available for this finding.