Back to skill

Security audit

iLovePDF

Security checks for vulnerabilities and agentic risk

Overview

This PDF helper needs review because it incorrectly treats password and document changes as safe reads and directs the agent to run an unverified remote installer.

Review before installing. Install a verified CLI yourself and review its automatic skill changes; the publisher documents OO_SKILLS_SYNC_DISABLED=1 to disable startup skill synchronization in its [command reference](https://github.com/oomol-lab/oo-cli/blob/main/docs/commands.md#environment-variables). Authorize PDF transformations and password changes explicitly, and submit only documents and PDF passwords you approve for processing through OOMOL/iLovePDF. Remote installers and live connector code were not audited.

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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (7)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The supplied code installs software from a different service and contains no iLovePDF search or read functionality. Nothing in the supplied chunk establishes this installation as a supporting implementation detail for the declared purpose. The downloaded script's further behavior is unknown.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
86% confidence
Finding

The supplied code performs only a login operation, rather than the declared iLovePDF functionality. Authentication could be a supporting implementation detail, but this chunk provides no evidence that the CLI or login targets iLovePDF or supports the described operations.

Content

No source excerpt is available for this finding.

External Script Fetching

High
Category
Supply Chain
Confidence
97% confidence
Finding

The installation command executes a mutable remote script without a pinned version or independent integrity verification; the PowerShell alternative has the same problem. Compromise of the distribution endpoint could therefore execute arbitrary code with the agent's privileges, although the command-not-found condition and any enforced oo-only tool restriction limit reachability.

Content

Scanner excerpt · SKILL.md (reported line 69)May include surrounding context.

The action workflow lacks a privacy warning about sending document inputs and PDF passwords to external services.

Medium
Category
Output Handling
Confidence
90% confidence
Finding

The workflow instructs the agent to execute connector actions with their input payloads; supported inputs include document URLs and, for unlocking PDFs, opening passwords (L53). Although the skill names OOMOL and iLovePDF, it does not warn that these inputs are transmitted externally and documents become accessible to the processing service; L58 instead broadly labels untagged actions as safe reads.

Content

No source excerpt is available for this finding.

The manifest describes searching and reading data, but the action catalog includes document-modification operations.

Medium
Category
Excessive Agency
Confidence
90% confidence
Finding

The read-oriented description understates a catalog that can transform documents and change protection on generated outputs. Combined with the instruction that all untagged actions are safe reads, this can cause an agent to exceed a read-only task's authorization, although the text does not establish that source files are overwritten.

Content

No source excerpt is available for this finding.

The safety instructions classify all untagged actions as reads, although untagged actions modify document content and protection.

Medium
Category
Excessive Agency
Confidence
98% confidence
Finding

None of the catalog actions carries a write or destructive tag, so the safety rule classifies document transformations and protection changes as reads. This creates an authorization gap that could permit unrequested document processing or security-property changes in generated outputs, without establishing deletion or modification of originals.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
80% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 90)May include surrounding context.

md
- **`OOMOL_INSUFFICIENT_CREDIT`** — OOMOL billing stop. Recharge at `https://console.oomol.com/billing/token-recharge` before retrying.

- **HTTP 402 without `OOMOL_INSUFFICIENT_CREDIT`** — the third-party account itself has no active plan or credit. Report the provider error message as is; do not ask the user to recharge OOMOL.

## Resources