Back to skill

Security audit

Pdf Poppler Utils

Security checks for vulnerabilities and agentic risk

Overview

This PDF utility skill mostly matches its stated purpose, but it includes under-scoped instructions that can change the user's Homebrew environment and exposes bypass/signing capabilities without enough guardrails.

Review this skill before installing. The Poppler tools themselves are common PDF utilities, but do not let an agent unlink Homebrew curl unless you have confirmed the exact conflict, understand the impact, and have a rollback plan. Use owner-password, DRM override, hidden-text extraction, attachment extraction, signing, and online certificate-fetching options only for PDFs you are authorized to process, and direct file-producing commands to an explicit output directory.

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 (1)

T09 · Insecure Skill Coding Practices

Warning
Location
SKILL.md:38
Finding

Unnecessary Global Modification of the Homebrew curl Environment

Content
View full analysis
⚠️ **Please ask for human permission before executing `brew unlink curl -v`.** If the human is unsure, recommend executing it (to resolve the potential conflict). ``` ### Technical Analysis The Skill is intended to perform PDF inspection, extraction, conversion, merging, splitting, attachment management, and signature verification through Poppler utilities. None of these functions inherently requires unlinking the globally installed Homebrew version of `curl`. The documented command changes Homebrew's global executable links and can consequently alter which `curl` binary is selected by the shell. Although the instructions require human permission, they also direct the agent to recommend execution when the user is uncertain. This encourages a system-wide configuration change without first confirming that a conflict exists. This exceeds the minimum system modification required for the declared PDF functionality. It may also create inconsistent TLS, certificate, or feature behavior if unrelated applications subsequently resolve a different `curl` implementation. ### Attack Path 1. A user loads the Skill to perform an ordinary local PDF operation. 2. The Skill recommends running `brew unlink curl -v` before Poppler is used. 3. The user grants permission without an actual curl conflict having been diagnosed, potentially because the Skill r ...[truncated 1480 chars]
Remediation
View remediation
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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (14)

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The skill tells the agent to recommend a specific system-changing action even when the human is unsure, which biases the agent away from preserving user choice and informed consent. Because brew unlink curl -v changes the local toolchain and can disrupt other software, this instruction can lead to unnecessary or harmful environment modifications without adequate validation.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The documentation exposes an option described as an owner password that 'bypasses all security' without any caution or limitation, which can normalize or encourage use of a capability that defeats PDF permission controls. In an agent context, this increases the risk that automated workflows use owner-password functionality to alter protected PDFs without clear user awareness or authorization checks.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The documentation includes the owner-password option and even notes that it 'bypasses all security' without any safety guidance. In an agent workflow, this can normalize bypassing document restrictions and cause the system to process or extract protected attachments contrary to policy, authorization, or user expectations.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The skill documentation states that pdfinfo prints metadata and can print all JavaScript, URLs, and document structure from a PDF, which may expose private or sensitive information from user documents. The markdown does not include any warning about handling confidential PDFs or reviewing output before sharing or logging it.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The -opw password option explicitly notes that it bypasses PDF security, which is a sensitive capability affecting protected documents. The file documents this behavior but provides no caution about authorization, privacy, or responsible use.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The documentation frames pdfsig as a verification-only tool, but it also exposes state-changing and side-effecting capabilities such as adding signatures, signing documents, dumping signatures to disk, and fetching certificates online. This mismatch can cause an agent or user to invoke the tool under false assumptions, leading to unintended file modification, data export, or network access.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The markdown lists -dump, -add-signature, and -sign without warning that they can write files to disk or modify the input/output PDF workflow. In an agent setting, omission of these side effects increases the risk of accidental data disclosure, overwriting artifacts, or unauthorized document changes when the tool is selected based on incomplete documentation.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The documentation exposes -nodrm as a routine conversion option without any warning or access-control context, which normalizes bypassing PDF usage restrictions. In an agent skill, this can enable unauthorized extraction or transformation of protected documents, going beyond ordinary PDF-to-HTML conversion.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill is described as a general PDF manipulation/conversion tool, but the DRM override capability is not necessary for the stated purpose and increases misuse potential. Including it without justification broadens the tool's scope into potentially unauthorized access to protected content.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The documentation mentions both DRM override and hidden text extraction as ordinary options but provides no safety, legal, or privacy warnings. In an automated agent context, that omission increases the likelihood that sensitive, intentionally concealed, or access-restricted content will be extracted without informed user intent.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
81% confidence
Finding

The skill documents commands that create new files or write outputs, but it does not clearly warn that output paths may create or overwrite files depending on user choice and tool behavior. In an agent setting, omission of file-write cautions increases the chance of accidental data loss, clobbering prior outputs, or unintended modifications in the working directory.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
84% confidence
Finding

This is a markdown file, so missing-warning checks apply to documented behaviors that could affect user data or system integrity. The description and options explain attachment operations, including replacing an embedded file and producing an output PDF, but they do not explicitly warn users that the command changes document contents and may overwrite prior embedded content.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
87% confidence
Finding

The skill documents extracting embedded files from PDFs but does not warn that extraction writes attacker-controlled filenames and content to disk. In an agent context, this can lead to unintended local file creation, accidental overwrite of user data, or dropping untrusted files for later execution or exfiltration.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Low
Category
Not specified by scanner
Confidence
78% confidence
Finding

The -aia option enables network retrieval of missing certificates, which expands the tool from local PDF processing into external communication. In an agent skill advertised as local PDF manipulation, undocumented or under-justified network behavior can violate isolation expectations and potentially leak metadata about processed documents or the environment.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.