Security audit

Notification System

Security checks for vulnerabilities and agentic risk

Overview

This is a legitimate notification skill, but it needs Review because it can send real outbound messages and has weak safeguards around recipients, rate-limit bypass, logs, and links.

Review before installing in any environment that can actually send messages. Replace the example recipients with placeholders, require explicit operator confirmation for sends and broadcasts, validate or generate all links server-side, enforce consent and opt-outs, protect or minimize delivery logs, and avoid letting callers choose high priority without authorization and bounded quotas.

A.I.G

Vulnerability Patterns

Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
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

Findings (3)

T09 · Insecure Skill Coding Practices

Error

Location: SKILL.md:166
Finding: Caller-Controlled Priority Can Bypass Notification Rate Limiting
Content: View full analysis
Remediation: View remediation

T09 · Insecure Skill Coding Practices

Error

Location: templates/email/invoice.md:57
Finding: Unvalidated Template URLs Enable Phishing and Unsafe Link Injection
Content: View full analysis
Pay Now
{{/if}} ``` `templates/email/invoice.md:67-70`: ```html

To unsubscribe from these emails, please contact us or click here.
``` `templates/email/notification.md:17-20`: ```html {{#if cta_url}}
View Details
{{/if}} ``` `templates/email/notification.md:25-28`: ```html

To unsubscribe, click here.
``` `templates/email/welcome.md:25-28`: ```html {{#if cta_url}}
Get Started
{{/if}} ``` `templates/email/welcome.md:35-38`: ```html

To unsubscribe from these emails, click here.
``` `templates/telegram/broadcast.md:1 ...[truncated 2492 chars]
Remediation: View remediation

T09 · Insecure Skill Coding Practices

Note

Location: SKILL.md:58
Finding: Executable Examples Contain Hard-Coded Real-Looking Recipient Identifiers
Content: View full analysis
Remediation: View remediation
`, ``, and ``. - If a numeric example is required, use an officially reserved fictional number and clearly label it as non-routable. - Require explicit recipient confirmation before the agent performs an outbound send. - Separate preview and execution steps, making preview the default. - Require an explicit confirmation flag for immediate delivery. - Validate that production recipients originate from an approved contact store rather than documentation examples. - Add tests or linting rules that reject realistic phone numbers, chat identifiers, and email addresses in executable examples. ]]>

SkillSpector

Vulnerability Patterns

Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (4)

Missing User Warnings

Medium

Category: Not specified by scanner
Confidence: 89% confidence
Finding: This markdown file describes immediate and scheduled sending of notifications to external channels like WhatsApp and Telegram, including explicit recipient targets. It also documents delivery logs containing channel and target data, but provides no warning that using the skill transmits recipient/system data to third-party services and stores identifying contact information in local logs.
Content: No source excerpt is available for this finding.

Missing User Warnings

Medium

Category: Not specified by scanner
Confidence: 89% confidence
Finding: The skill operationalizes outbound messaging across multiple channels and stores queue and delivery data, including recipient identifiers and message content variables, but it provides no privacy, consent, retention, or access-control guidance. In this context, that omission can lead agents or operators to process personal data, send messages without adequate authorization, and retain sensitive metadata in logs, increasing compliance and data exposure risk.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

Category: Not specified by scanner
Confidence: 88% confidence
Finding: This file applies quiet hours using the fixed timezone "America/Los_Angeles", which imposes a locale-specific behavior on all users unless overridden elsewhere. Under the policy, forcing a specific locale without user opt-in is a natural-language policy concern even in configuration files.
Content: No source excerpt is available for this finding.

Missing User Warnings

Medium

Category: Not specified by scanner
Confidence: 82% confidence
Finding: This manifest advertises promotional broadcast templates such as "Broadcast Promotion" but provides no disclosure in the file about the user-impacting nature of sending marketing content. For markdown and manifest-style descriptions, safety-sensitive behaviors that can affect privacy or user communications should be accompanied by clear warnings or usage constraints.
Content: No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.