Inbox

• Outcome: Review
• Latest audit: Sep 10, 2026 · 8:21 PM UTC
• Version: 1.0.0

Overview

The skill is mostly an inbox advice guide, but some files instruct automatic archiving, sending, declining calendar invites, transcription, and broad channel aggregation without clear user approval boundaries.

Review this skill carefully before installing if your agent has email, calendar, chat, or messaging tools. Use it in draft-and-review mode unless you explicitly configure automation, because some guidance could otherwise hide messages or send communications on your behalf.

A.I.G

Vulnerability Patterns

• Skill Instruction Hijacking: Alters the agent's session goals or safety constraints when the skill loads
• Agent Memory Poisoning: Writes attacker-controlled rules into memory that affect later sessions
• Remote Payload Retrieval and Execution: Fetches external code whose behavior can change after review
• Embedded Malicious Code: Ships malicious scripts inside the skill and executes them locally
• Unauthorized Access and Privilege Escalation: Obtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (9)

Vague Triggers

Medium

• Confidence: 85% confidence
• Finding: The 'When to Use' section says the skill applies when a user needs help managing incoming streams across many channels and 'any inbox type,' which is a very broad activation description. While the scope section limits capabilities, it does not provide specific trigger phrases, exclusion examples, or clearer boundaries for when this skill should versus should not be invoked.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The skill explicitly promotes aggregating messages and notifications from many personal and work channels into a single view, but provides no warning about the breadth of data access, commingling of sensitive contexts, or privacy implications. This creates a real risk of over-collection, accidental exposure of confidential communications, and user misunderstanding about how much content the agent will process.

Missing User Warnings

Medium

• Confidence: 88% confidence
• Finding: The email handling guidance includes automated actions such as auto-archiving newsletters without warning that these actions can alter the user's communications state and potentially hide important messages. Even if intended for productivity, silent automation over inbox content can cause missed obligations, lost evidence, or suppression of messages the user expected to review.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: Automatically declining calendar events with a polite response has direct external side effects because it sends communications to other people and can affect scheduling, business relationships, or commitments. Without an explicit warning and approval model, the agent could unintentionally reject important meetings or send inappropriate responses on the user's behalf.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: Immediate transcription of voice messages processes potentially sensitive spoken content, including personal data, credentials, health information, or confidential business details, yet no privacy warning or consent boundary is described. This is especially risky because voice content may come from third parties who did not expect automated transcription and summarization.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: Automated follow-up nudges with escalating tone authorize outbound messaging behavior that can affect external relationships, create harassment-like patterns, or send inaccurate pressure signals without user review. In the context of multi-channel orchestration, this is more dangerous because the automation may act across email, chat, and other platforms at scale.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: The skill recommends auto-archiving older items 'out of sight' after extracting only a subset of messages, but it does not explicitly require a clear user warning, confirmation, or an easy review path before bulk action. In an inbox-management context, this can cause users to miss important communications that were incorrectly excluded from the extraction criteria, creating a real risk of data-handling mistakes and loss of visibility.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The workflow explicitly allows "Pre-approved auto-send" with a "Fully automated" automation level for outbound communications, but it provides no user-facing warning, approval checkpoint, or constraint around what may be sent on the user's behalf. In an inbox-management skill, this is risky because incorrect tone, bad recipients, hallucinated details, or unauthorized commitments could be transmitted directly to third parties without human review.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: The skill instructs the agent to classify items into a 'Noise' bucket and 'Auto-archive' them before showing anything to the user, but it provides no safeguard, disclosure, or review mechanism. In an inbox-management context, this can hide legitimate messages from the primary view, causing missed deadlines, lost requests, or suppressed communications based on imperfect automated triage.

Static analysis

No suspicious patterns detected.