Back to skill

Security audit

GPT Image Edit — Pro Pack on RunComfy

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed RunComfy image-editing helper, with ordinary third-party CLI and token-handling risks users should understand before installing.

Install only if you are comfortable sending prompts and public image URLs to RunComfy for hosted processing. Prefer a pinned or isolated CLI install, avoid running npm install with elevated privileges, and use a limited RUNCOMFY_TOKEN or account where possible.

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:40
Finding

Unpinned Global Installation of a Third-Party CLI

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, line 40
Vulnerability Type: Unpinned third-party dependency installed globally
Risk Level: Medium

Complete Code Snippet:

bash
npm i -g @runcomfy/cli

Technical Analysis

The documented prerequisite installs the latest available version of @runcomfy/cli globally without an exact version pin, lockfile, or integrity verification. Consequently, the package contents executed by users can change after the Skill has been reviewed.

npm installation may execute package lifecycle scripts with the privileges of the invoking user. If the package, publisher account, or distribution channel is compromised, following this instruction could execute attacker-controlled code locally. The global installation scope also makes the installed executable available outside this Skill and increases the potential impact of a compromised or incompatible release.

The CLI's network communication with RunComfy is necessary for the declared hosted image-editing functionality and is explicitly disclosed. The identified issue is therefore not covert exfiltration, but the unsafe acquisition and global installation of an uncontrolled dependency version.

Attack Path

  1. An attacker compromises the @runcomfy/cli npm package, its publisher account, or a newly published package version.
  2. The attacker adds malicious installation lifecycle code or modifies the CLI's runtime behavior.
  3. A user follows the prerequisite in SKILL.md and runs npm i -g @runcomfy/cli.
  4. npm retrieves the current uncontrolled release and may execute its lifecycle scripts with the user's privileges.
  5. The malicious package can access resources available to that user, including user-readable files, environment variables, RunComfy credentials, and files supplied to subsequent CLI operations.
  6. Because the package is installed globally, the compromised executable remains available for later invoc ...[truncated 745 chars]
Remediation
View remediation

Remediation Suggestions

  1. Pin @runcomfy/cli to an exact, reviewed version rather than installing the latest release, for example:
    bash
    npm install --global @runcomfy/cli@<reviewed-exact-version>
    
  2. Publish and verify the expected package version, npm registry, publisher identity, and integrity hash in the installation documentation.
  3. Prefer a project-local dependency governed by a committed lockfile instead of a global installation where operationally feasible.
  4. Use deterministic installation mechanisms such as npm ci with a reviewed lockfile.
  5. Disable npm lifecycle scripts during installation when the package supports operation without them, and separately perform any strictly necessary trusted setup.
  6. Run installation and subsequent CLI operations as an unprivileged user with access limited to the files and environment variables required for the image-editing task.
  7. Review dependency updates before changing the pinned version and monitor the package and publisher for supply-chain compromise.
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
94% confidence
Finding

The trigger phrase 'any explicit ask to edit with this model' is overly broad and can cause the skill to activate in situations where the user did not clearly intend to invoke this specific local CLI-backed integration. In an agent setting, broad matching increases the chance of accidental tool routing, unintended transmission of user-provided image URLs and prompts to a third-party service, and confusion with sibling image-edit skills that may have different privacy or behavior characteristics.

Content

No source excerpt is available for this finding.