Security audit

ComPDF Page Editor

Security checks across malware telemetry and agentic risk

Overview

This PDF page-editing skill has a reasonable main purpose, but it bundles much broader ComPDF API material while also using a local API key and external document uploads.

Review before installing. Use this only for ComPDF page-management tasks, and confirm that your agent will not use the bundled conversion, OCR, AI extraction, encryption/decryption, or other unrelated endpoints. Do not use it on sensitive PDFs unless you are comfortable uploading them to ComPDF, and keep the API key file private.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
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

Findings (8)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 92% confidence
Finding: The skill instructs the agent to read local API key files and references a maintainer workflow that writes refreshed documentation snapshots, yet it declares no permissions. Hidden or undeclared file and network capabilities reduce transparency and can bypass a user's expectations about what the skill is allowed to access or transmit. In this context, the risk is elevated because the skill also handles a secret API key and may communicate with external ComPDF infrastructure.

Tp4

High

Category: MCP Tool Poisoning
Confidence: 88% confidence
Finding: The stated purpose is limited to page-management operations, but the skill content and maintainer instructions indicate broader behavior: fetching remote documentation, parsing content, and generating local reference files. This description-behavior mismatch can mislead reviewers and users about the real trust boundary, making it easier for network/file actions to occur without informed consent. The danger is higher here because the skill also directs use of a locally stored API key and external service interaction.

Description-Behavior Mismatch

High

Confidence: 96% confidence
Finding: The reference file exposes a much broader capability surface than the skill's declared purpose of PDF page editing, including conversion, OCR, AI extraction, encryption, and document-generation operations. In an agent setting, overbroad in-scope documentation can cause tool misuse or prompt-induced invocation of unintended high-risk endpoints, violating least privilege and user expectations.

Context-Inappropriate Capability

High

Confidence: 97% confidence
Finding: Including AI parsing and extraction APIs in a page-editor skill expands the skill from structural PDF manipulation into full document-content extraction. That creates a materially different privacy and data-exfiltration risk, because sensitive uploaded documents could be parsed and extracted without users expecting content-level processing.

Context-Inappropriate Capability

Medium

Confidence: 87% confidence
Finding: Encryption and especially decryption capabilities are outside the stated page-management scope and enable transformation of document security state rather than page structure. In a mis-scoped skill, this can lead to unauthorized password removal or security-policy bypass on user-provided documents.

Context-Inappropriate Capability

Medium

Confidence: 90% confidence
Finding: Watermarking, compression, standards conversion, generation, and document comparison extend the skill into content transformation and analysis capabilities not implied by 'page editor'. This broader surface increases the chance of unintended destructive edits, hidden data handling, or invocation of features users did not authorize.

Context-Inappropriate Capability

High

Confidence: 98% confidence
Finding: Broad file-conversion and OCR workflows are unrelated to page editing and significantly increase the skill's functional scope into full document ingestion and content extraction. This is dangerous because OCR and conversion can expose document contents to third-party processing and can be abused to perform covert data transformation beyond the user's requested task.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The documentation describes uploading files to ComPDF and receiving download URLs, but it does not clearly warn users that their documents and potentially extracted contents are sent to an external service. This omission undermines informed consent and can cause sensitive data to be transmitted off-platform unexpectedly.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.