Back to skill

Security audit

paper-parser-skill

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed arXiv paper parsing helper, but users should verify the external PyPI package and avoid sending sensitive PDFs to MinerU.

Install in a dedicated virtual environment or container, review the referenced source and exact package version before use, protect the MinerU token, and do not submit confidential or unpublished PDFs unless you accept MinerU processing them.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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)

T08 · Insecure Dependencies

Warning
Location
SKILL.md:38
Finding

Unverified Third-Party Package Installation

Content
View full analysis

Vulnerability Details

File Location: SKILL.md:38-45
Vulnerability Type: Insecure third-party dependency installation
Risk Level: Medium

Vulnerable Code

markdown
## 🚀 Setup

> [!WARNING]
> Installing from PyPI executes third-party code. Use a virtual environment if you want to limit blast radius.

```bash
pip install paper-parser-skill==v0.1.3
text

### Technical Analysis

The setup instructions direct users to retrieve and install an executable Python package from PyPI. Although the package version is pinned, the instructions do not provide or enforce cryptographic hashes, signed provenance, a verified lockfile, or another integrity control. The reviewed project contains only `SKILL.md`, so the code executed by this command is not included in the audit scope.

Version pinning prevents automatic selection of a newer release but does not independently establish that the downloaded distribution is authentic or identical to a previously reviewed artifact. The Skill metadata also declares version `0.1.4`, while the installation command selects package version `v0.1.3`, creating additional ambiguity about which implementation corresponds to the reviewed Skill documentation.

This is a supply-chain risk rather than evidence that the referenced package is currently malicious. Exploitation depends on compromise, replacement, or misrepresentation of the external package or one of its transitive dependencies.

### Attack Path

1. An attacker compromises the referenced PyPI distribution, its publisher account, build process, or a transitive dependency.
2. The attacker introduces malicious installation or runtime code into an artifact retrieved through the documented command.
3. A user follows the Skill instructions and runs `pip install paper-parser-skill==v0.1.3`.
4. Package installation or subsequent CLI execution runs the attacker-controlled code with the user's privileges.
5. The malicious c
...[truncated 866 chars]
Remediation
View remediation

Remediation Suggestions

  1. Publish a reviewed dependency lockfile containing cryptographic hashes for the package and every transitive dependency.
  2. Require hash verification during installation, such as with pip install --require-hashes -r requirements.txt.
  3. Provide verifiable package provenance or signatures and document how users should validate them before installation.
  4. Bundle the reviewed source code with the Skill or pin it to an immutable, reviewed commit so the executable implementation can be audited alongside the documentation.
  5. Align the Skill metadata version with the package version referenced by the installation instructions.
  6. Recommend installation into a dedicated virtual environment or restricted container under a non-privileged account.
  7. Avoid installing with sudo or into a system Python environment.
  8. Use a dedicated, minimally scoped, revocable MinerU token and restrict access permissions on ~/.paper-parser/config.yaml.
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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
Findings (1)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The trigger list includes generic terms like "paper," "search," and "download," which are common in many unrelated contexts and can cause the skill to activate when not specifically intended. In an agent setting, overbroad activation can route user requests into a tool that downloads files and sends PDFs to an external service, creating unnecessary execution and data exposure risk.

Content

No source excerpt is available for this finding.