Back to skill

Security audit

ConvertAPI

Security checks for vulnerabilities and agentic risk

Overview

The conversion workflow is coherent, but its automatic setup executes an unverified remote installer and under-discloses broader changes the CLI can make.

Review and install the CLI separately before enabling this skill, including its automatic agent-skill synchronization settings. The remote installers and hosted connector implementation were not verified. Use a PDF you intend to process through OOMOL and ConvertAPI, and consider service charges. The review found no artifact-backed evidence of malicious intent.

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 (6)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied command executes a remote installer from a different service without demonstrating any ConvertAPI functionality. This is a materially different action from the declared searching and reading purpose. The installer contents are not supplied, so its subsequent behavior cannot be determined.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
90% confidence
Finding

The supplied code only initiates a CLI login; it does not perform any identifiable ConvertAPI operation. Authentication could be a supporting implementation detail, but nothing in this chunk establishes that oo authenticates to ConvertAPI or supports the declared functionality.

Content

No source excerpt is available for this finding.

External Script Fetching

High
Category
Supply Chain
Confidence
98% confidence
Finding

Piping a live HTTPS response directly into bash executes installer content without a pinned version or independently verified integrity. A compromised distribution endpoint or altered installer could execute arbitrary code with the invoking user's privileges; HTTPS alone does not protect against compromise of the publisher's infrastructure.

Content

Scanner excerpt · SKILL.md (reported line 55)May include surrounding context.

The conversion action is untagged and therefore incorrectly classified as a read-only operation.

Medium
Category
Excessive Agency
Confidence
94% confidence
Finding

The conversion action creates a derived document and temporary download URLs, but the blanket rule classifies untagged actions as reads that are safe to execute directly. Conversion can also consume service credits and involve third-party document processing, so this classification can bypass the skill's own confirmation requirement for consequential operations.

Content

No source excerpt is available for this finding.

The setup workflow executes remotely downloaded installation scripts without warning the user about the system changes or obtaining approval.

Medium
Category
Output Handling
Confidence
96% confidence
Finding

When the CLI is missing, the setup instructions direct execution of downloaded shell or PowerShell code without explicitly obtaining approval for software installation. If followed, this can modify the local environment and execute code with the invoking account's privileges; the conditional setup trigger limits exposure but does not establish consent.

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 76)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