Back to skill

Security audit

Learn Anything

Security checks for vulnerabilities and agentic risk

Overview

This learning skill is purpose-aligned and mostly transparent, with some privacy considerations around optional progress tracking and exports.

Before installing, treat progress journals and exports as personal data. Confirm where files will be written and approve each export to external tools, especially Notion or calendars. Also note that the Anki/export command examples may need correction before they work.

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
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The trigger list is very broad and matches common conversational requests like 'help me learn X' or 'teach me X', which increases the chance of unintentional activation. In an agent environment, that can cause the skill to take over ordinary educational queries unexpectedly and steer behavior, resource selection, or file-writing flows without the user clearly opting in.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The quick-start examples reinforce the same ambiguous activation model by presenting ordinary user utterances as activation commands. This makes accidental invocation more likely and can blur the distinction between casual Q&A and a workflow that may create artifacts, track progress, or influence follow-on actions.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The workflow says 'Track progress — Update learning journal' without warning that study activity may be persisted to a local file. Users may disclose sensitive interests, goals, schedules, or performance data, and silent persistence creates privacy and consent risks even if the content is educational.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The integration/export section implies creating files and updating external tools such as Anki, Obsidian, Notion, and calendars, but it does not state what data is written where or whether consent is required. This can lead to unexpected persistence or propagation of user learning data into third-party systems, increasing privacy exposure and potential data leakage.

Content

No source excerpt is available for this finding.