ClawHub
Back to skill

Security audit

Math Modeling Contest Route Selection

Security checks for vulnerabilities and agentic risk

Overview

This skill is a coherent contest-planning helper that reads local guidance files and optionally runs a local scoring script, with no evidence of hidden access, persistence, exfiltration, or destructive behavior.

Before installing, understand that the skill may guide your agent to run a bundled Python scoring script on contest-ranking JSON and may write a markdown report if an output path is provided. Use it in workspaces where creating contest-planning files is acceptable.

A.I.G

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

SkillSpector

Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (4)

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
93% confidence
Finding

The skill instructs the agent to read multiple reference files and run a local script, which implies file read and possible file write/code-execution capabilities, but it does not declare an explicit tool scope such as permissions or allowed-tools. That creates an authorization gap: a host agent may grant broader filesystem access than the skill actually needs, increasing the chance of unintended file access, data exposure, or side effects if the skill is invoked in a sensitive workspace.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · references/method-map.md (reported line 59)May include surrounding context.

md
- Metaheuristic without a baseline or convergence evidence.
- Fuzzy evaluation used to hide unclear indicators.
- Regression treated as causal proof.
- Raw Euclidean clustering or PCA on percentage-sum composition data without checking closed-sum effects.
- Bearing-only localization without identifiability, mirror-symmetry, or error-propagation checks.
- Stacking three advanced methods when one baseline plus one justified upgrade would be clearer.
- Optimizing an objective that is not connected to the deliverable asked by the problem.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
94% confidence
Finding

The description explicitly states that, aside from the English main title, the rest of the banner is presented in Chinese. This is a natural-language locale constraint in the asset with no indication of user opt-in or justification that the skill is intended only for a Chinese-language audience.

Content

No source excerpt is available for this finding.

Vague Triggers

Low
Category
Not specified by scanner
Confidence
81% confidence
Finding

The phrase describing use of the file includes 'when the user asks for multi-dimensional scoring standards based on mathematical modeling evaluation practice,' which is fairly broad and does not define clear trigger boundaries or exclusions. In a markdown skill/reference file, this ambiguity could cause the material to be invoked for loosely related scoring or evaluation requests beyond the intended paper-competitiveness context.

Content

No source excerpt is available for this finding.