Versioned security assessment

Report ID: SA-EB5CFDF2

7/1/2026, 2:46:27 AM

mcp-builder security assessment v2

Skill Security Certification Report

Audit History

Audit model: codex Historical report

Skill name: mcp-builder
Version: v2
Maintainer: ZhanlinCui
Source: github.com/ZhanlinCui/Ultimate-Agent-Skills-Collection/tree/main/mcp-builder
Coverage: 9 Files scanned · 3,330 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 produced many severe-looking hits, but review shows most are Markdown examples, MCP terminology, placeholder URLs, and normal SDK guidance. The real risks are contextual: the optional evaluation harness can run user-selected stdio commands, connect to remote MCP endpoints, and pass credentials supplied by the user. No prompt injection, obfuscation, hardcoded live secret, 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

9 Files scanned · 3,330 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.

Observed in 2 evidence locations

Network access

May connect to external services.

Observed in 4 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 5 evidence locations

External commands

May invoke commands or programs outside the Skill.

Observed in 3 evidence locations

Risk findings

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

Confirmed security concerns (2)

RISK-001 Medium

scripts/connections.py:76-84 scripts/evaluation.py:327-353

User-selected MCP server command execution

The evaluation harness can start a stdio MCP server from a command and arguments supplied by the user. This is expected for testing local MCP servers, but it can execute untrusted code and pass environment variables to that process.

The harness explicitly accepts a command, args, and env values for stdio transport and passes them into the MCP stdio connection. The behavior is intentional for an evaluator, so the risk is contextual rather than malicious.

RISK-002 Medium

scripts/connections.py:91-109 scripts/evaluation.py:331-353 scripts/evaluation.py:315-318

User-supplied remote MCP endpoint connections

The evaluation tooling supports SSE and HTTP connections to user-provided URLs with optional headers. This is legitimate MCP test functionality, but users could expose bearer tokens or connect to untrusted endpoints if they supply unsafe values.

The code clearly creates remote MCP clients from caller-provided URLs and headers. The examples use placeholder endpoints, so there is no evidence of hardcoded exfiltration.

Needs review findings (1)

REVIEW-001 Low

reference/evaluation.md:395-399 reference/evaluation.md:553-568 reference/node_mcp_server.md:704-744 reference/python_mcp_server.md:506-517

Credential handling appears in examples and evaluator inputs

The skill documents API keys, tokens, and environment variable use for MCP testing and generated servers. The examples are placeholders and mostly recommend environment variables, but users should avoid passing real secrets to untrusted servers or remote endpoints.

The references show placeholder API keys and token-passing patterns, not live secrets. The concern remains because the evaluator and generated servers are designed to work with real credentials.

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 (1)

Low

SKILL.md:37-74 reference/mcp_best_practices.md:152-172 reference/python_mcp_server.md:68-119 reference/node_mcp_server.md:584-601

Most static analyzer hits are documentation false positives

The reported Ruby backtick and weak cryptography findings occur in Markdown guides and code examples, not in hidden execution logic. Reviewed content shows MCP implementation guidance, SDK examples, and security best practices rather than malicious obfuscation.

The reviewed locations are prose and illustrative snippets. I found no evidence of weak cryptographic routines, obfuscation, or Ruby shell backticks being used by the skill runtime.

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