Versioned security assessment

Report ID: SA-5AF94561

6/30/2026, 8:08:17 AM

chatgpt-app-builder security assessment v2

Skill Security Certification Report

Audit History

Audit model: codex Historical report

Skill name: chatgpt-app-builder
Version: v2
Maintainer: mcp-use
Source: github.com/mcp-use/mcp-use/tree/main/skills/chatgpt-app-builder/
Coverage: 18 Files scanned · 2,127 Lines analyzed
Policy version: Unavailable

Highest confirmed finding severity

Medium

2 confirmed security findings require attention.

Installation context

Historical evidence

This report may not describe the currently installable artifact. Open the current Skill page for install guidance.

Open current Skill page

This report does not block or authorize the manifest or ZIP.

Static analysis reported many high-severity patterns, but review found they are mostly Markdown examples, API names, code fences, and license text rather than executable malicious code. The confirmed risk is legitimate developer guidance that may run npm/npx commands, start tunnels, use external URLs, and read environment-driven configuration. No prompt injection, credential exfiltration, obfuscation, or confirmed malicious intent was found.

Report position

Historical report

Open audit history before using this report to install.

Audit attestation

Not attestable

The required immutable binding is incomplete.

Human verification

Not verified

No human verification is recorded for this report.

Coverage

18 Files scanned · 2,127 Lines analyzed

3 items shown for review

Limitations

This report does not claim runtime or sandbox execution and does not prove the absence of side effects.

Evidence chain

Follow the evidence from source binding to the install contract. Available evidence supports verification; it is not a safety guarantee.

Source

Binding unavailable
Artifact

Identity incomplete
Audit

Complete
Install contract

Open manifest to verify
Open manifest

Capabilities observed

Observed means this report recorded supporting evidence. Not recorded does not prove that a capability is absent.

Contains scripts

May execute code included with the Skill.

Not recorded by this audit

Network access

May connect to external services.

Observed in 5 evidence locations

Filesystem access

May read or write local files.

Not recorded by this audit

Env variables

May read values from the process environment.

Observed in 2 evidence locations

External commands

May invoke commands or programs outside the Skill.

Observed in 5 evidence locations

Risk findings

Confirmed security concerns are separated from items that still need review.

Confirmed security concerns (2)

RISK-001 Medium

SKILL.md:4-7 references/setup.md:7-10 references/setup.md:55-59 references/setup.md:81-84

Developer Command Execution Guidance

The skill instructs agents and users to install skills, scaffold projects, install packages, start tunnels, and deploy apps. This is expected for an app-builder skill, but commands can change the filesystem, install dependencies, and expose a local service if executed without review.

The command examples are explicit and semantically relevant to the skill. They are legitimate development commands, not evidence of malicious intent.

RISK-002 Medium

references/setup.md:36-42 references/setup.md:55-64 references/csp-and-metadata.md:22-32 references/components-api.md:275-284

External Network and Tunnel Configuration

The references guide users to local MCP endpoints, tunnels, external checkout links, and CSP allowlists for third-party domains. This is legitimate for ChatGPT apps, but it increases review requirements for network exposure and data sharing.

The network behavior is directly documented and expected for app integration. The reviewed files do not show credential exfiltration or hidden destinations.

Needs review findings (1)

REVIEW-001 Low

references/server-and-widgets.md:26-30 references/csp-and-metadata.md:70-72

Environment Configuration Examples

The skill shows environment-based URL configuration for MCP base URLs and optional CSP domains. This is common configuration guidance, but users should avoid placing secrets in generated client-visible metadata or logs.

The examples access environment-driven configuration, but only for URLs and CSP domains. No evidence found that secrets are read or transmitted.

Expert evidence

Immutable subject identity, scanner metadata, dismissed matches, and source-level evidence.

Artifact subject

Marketplace commit: Unavailable
Content hash: Unavailable
Tree hash: Unavailable
Skill path: Unavailable
Audit payload hash: Unavailable

Analysis metadata

Audit model: codex

Analysis state: Complete

Scope is limited to the recorded files, lines, methods, and evidence. No runtime or sandbox execution is claimed.

Static false positives ignored (3)

Low

evals/architecture.json:19-20 references/architecture.md:3 LICENSE.txt:25

Weak Cryptography Alerts Are False Positives

Static findings labeled weak cryptography occur in documentation, evaluation text, and license prose. The reviewed files contain no executable cryptographic algorithm selection or hashing implementation.

The cited locations are prose or eval expected behavior, not executable crypto code. No evidence found of MD5, SHA1, DES, or similar usage in runtime code.

Low

evals/README.md:9-16 references/server-and-widgets.md:49-58 references/state-and-context.md:233-235

Backtick Execution Alerts Are Markdown and JavaScript Examples

Many Ruby backtick alerts are Markdown fences, inline code, or JavaScript template strings inside documentation. They are not Ruby shell execution and are not packaged runtime code.

The reviewed locations are documentation snippets and fenced examples. No Ruby source file or shell backtick execution was found.

Low

references/components-api.md:172-189 references/ui-guidelines.md:139-151 references/ui-guidelines.md:168-190

System Reconnaissance Alerts Are Responsive UI Context

The system-reconnaissance findings correspond to widget host context such as device type, safe area, locale, and display mode. This supports responsive UI behavior and is documented as part of the widget API.

The context is limited to documented widget environment fields. No evidence found of OS probing, file enumeration, or unauthorized host reconnaissance.

Verify and export

JSON report CITATION.cff SARIF 2.1.0 Signed installation manifest

The manifest and lockfile bind install artifacts to cryptographic hashes. This integrity claim is separate from the security assessment.

Audit attestation: not_attestable