Back to skill

Security audit

market-data

Security checks for vulnerabilities and agentic risk

Overview

The skill describes a stock-market data helper but also includes a generic local learning script that stores usage notes and preferences outside the market-data purpose.

Review this before installing if you do not want a skill to keep local records of usage, notes, preferences, or errors. The market-data guidance itself is not destructive and no exfiltration was found, but the learner should ideally be removed, made opt-in, or confined to a documented per-skill data directory.

Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • 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
Findings (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
88% confidence
Finding
The skill includes commands for a "self-evolution learning system" that writes persistent local state to learned_patterns.json, but it declares no permissions. Undeclared file-write behavior weakens containment and user awareness, and if the learner is invoked broadly it could modify files or accumulate sensitive operational data without an explicit trust boundary.

Tp4

High
Category
MCP Tool Poisoning
Confidence
92% confidence
Finding
The advertised purpose is market-data retrieval, but the skill also instructs use of a generic learner that records history, failures, notes, preferences, insights, and reflections for arbitrary skills. This hidden stateful behavior expands the data-collection and persistence surface beyond user expectations, creating risks of unintended data retention, cross-skill profiling, and misuse of local files.

Description-Behavior Mismatch

Medium
Confidence
93% confidence
Finding
The file defines a generic cross-skill 'self-evolution' module that is explicitly reusable by any WorkBuddy skill, which exceeds the stated purpose of a market-data skill. This kind of scope expansion creates an unnecessary persistence and behavior-tracking capability that can be repurposed across unrelated skills, increasing privacy and governance risk even if no overtly malicious action is present.

Context-Inappropriate Capability

Medium
Confidence
96% confidence
Finding
The code persistently stores user preferences, operational history, error notes, and recent actions in learned_patterns.json without any consent flow, retention control, or access restriction. In the context of a market-data skill, this is unrelated to the declared purpose and can expose sensitive usage patterns or notes, especially because the skill can also resolve paths under ~/.workbuddy/skills and write files there.

Static analysis

No suspicious patterns detected.