Back to skill

Security audit

fuel-economy-driving-coach

Security checks for vulnerabilities and agentic risk

Overview

This local fuel calculator has no evident harmful access or hidden behavior, although some advertised calculations are overstated.

Treat results as rough estimates, not engine-specific analysis or verified payback. Check input units: the documented --frontal 24 example conflicts with the script's square-metre input and 2.2 default. Keep speed choices within safe, legal limits and follow vehicle-manufacturer guidance.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (4)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The code largely serves the declared fuel-saving purpose, but two explicit analytical capabilities are materially overstated. Its floor is a simplified road-load estimate rather than an engine-specific BSFC calculation, and its ranking measures annual savings rather than payback. The audit also uses a preset 75-versus-65-mph comparison rather than the advertised 75-versus-60 comparison, although the trip command supports custom speeds. No network access, file access, subprocess execution, or unrelated resource use appears in the supplied code.

Content

No source excerpt is available for this finding.

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.

The audit ranks estimated annual fuel costs rather than fixes by payback.

Low
Category
Excessive Agency
Confidence
98% confidence
Finding

The manifest promises to rank fixes by payback, but the audit sorts habits solely by the midpoint of their estimated annual fuel costs. It does not account for implementation costs or calculate payback periods, so its ordering is not the advertised return-on-investment ranking.

Content

No source excerpt is available for this finding.

The claimed engine-specific BSFC floor is actually calculated using a fixed 22% efficiency.

Low
Category
Excessive Agency
Confidence
99% confidence
Finding

The manifest promises a theoretical floor based on the user's engine BSFC. The implementation instead divides estimated road-load power by a hard-coded 0.22 efficiency, without accepting or calculating engine-specific BSFC, so the result does not provide the advertised engine-specific analysis.

Content

No source excerpt is available for this finding.