Security audit

Crypto Alert Aggregator

Security checks for vulnerabilities and agentic risk

Overview

The skill is not overtly malicious, but it presents random mock data as real crypto and Twitter alerts while reading credentials it does not actually use.

Install only for demo or development use unless the publisher clearly labels the data as synthetic or implements real, verifiable API retrieval. Do not provide real crypto, Twitter, or webhook credentials to this version, and do not rely on its alerts for trading decisions.

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

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
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
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection

Findings (7)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 93% confidence
Finding: The skill references environment variables for API keys and a webhook endpoint but does not declare any explicit tool scope or permission boundaries. That omission weakens reviewability and can lead to overbroad runtime access or unintended secret use, especially in agent environments where env access is sensitive.

Vague Triggers

Medium

Confidence: 95% confidence
Finding: The description is broadly phrased as generally useful for traders and researchers without clear invocation boundaries, exclusions, or safety constraints. In agent routing systems, vague trigger language can cause the skill to be invoked in contexts the user did not intend, increasing the chance of unnecessary external-data handling or alert dissemination.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The skill documents an external webhook for notifications but gives no warning that alert contents may be transmitted off-platform to third-party infrastructure. This can result in unintended data egress, leakage of sensitive trading interests or monitored assets, and surprise exposure of metadata to external services.

Description-Behavior Mismatch

Medium

Confidence: 97% confidence
Finding: The skill advertises real-time crypto market and Twitter signal aggregation, but the implementation only returns random mock data and never performs real external data collection. In a trading and research context, this is dangerous because users may make financial decisions based on fabricated alerts that appear legitimate.

Vague Triggers

Medium

Confidence: 91% confidence
Finding: The description uses broad activation wording ('Aggregates crypto market data and Twitter signals into actionable alerts for traders and researchers') without clear trigger boundaries or user-intent constraints. In agent ecosystems, vague descriptions can cause over-invocation or unintended selection in contexts involving trading, market monitoring, or social-signal analysis, increasing the chance the skill is used when not explicitly requested.

Context-Inappropriate Capability

Low

Confidence: 80% confidence
Finding: The code reads sensitive environment variables for crypto and Twitter API keys plus a webhook URL even though the current implementation does not use them. Unnecessary access to secrets increases the attack surface and creates unjustified privilege, especially in agent environments where skills should follow least-privilege design.

Missing User Warnings

Low

Confidence: 88% confidence
Finding: The skill accesses sensitive environment credentials without any disclosure to the user, despite the visible behavior being only mock-data generation. In an agent skill context, silent secret access is risky because users cannot tell that credentials are being consumed, and such access is not justified by the current functionality.

Static analysis

No suspicious patterns detected.