Skills create-mcp-skill
📦
create-mcp-skill
Content revision r1 High Risk ⚙️ External commands🌐 Network access📁 Filesystem access
Create MCP Skill Templates
Building a skill around an MCP server requires reliable discovery, testing, and documentation. This skill gives Claude, Codex, and Claude Code a structured workflow for MCP skill creation.
Supports: Claude Codex Code(CC)
Stop for confirmation before installing.
Review the plan and obtain explicit user consent before changing files.
Install with my Agent
Copy this request to your Agent. It includes the canonical Skill page and manifest.
Agent request
Review the Skillstore skill "create-mcp-skill" from https://skillstore.io/skills/cygnusfear-create-mcp-skill.md and its manifest at https://skillstore.io/api/skills/cygnusfear-create-mcp-skill/manifest. Verify the artifact. Stop and obtain explicit user consent before installing or changing files.Your Agent should still show its plan and request any confirmation required by the security policy.
Agent-readable resources
Use these links when an AI agent, crawler, or script needs clean context instead of reading the full page.
Test it
Using "create-mcp-skill". Create a skill for a browser automation MCP server.
Expected outcome:
- A skill folder plan with a clear SKILL.md outline.
- A list of discovered browser tools and safe test tasks.
- A quick start section with reviewed command patterns.
Using "create-mcp-skill". Improve an existing MCP skill draft.
Expected outcome:
- A revised structure for setup, quick starts, tool reference, and troubleshooting.
- Notes about unsafe commands, broad permissions, and missing package pinning.
- A checklist for testing examples before publication.
Using "create-mcp-skill". Prepare documentation after MCP tool discovery.
Expected outcome:
- Plain-language tool descriptions with parameter notes.
- Warnings for optional parameters, timeouts, and server-specific quirks.
- Practical next steps for completing the skill file.
Security Audit
High Riskv8 • 7/5/2026 Open versioned report
Most static findings are Markdown code fences, inline examples, or documentation links rather than literal Ruby backtick execution, path traversal, or hardcoded network callbacks. Context review still found high-risk guidance: the skill encourages execution of user-supplied MCP server commands, unpinned package runners, and a broad filesystem server example. No prompt injection text was found in SKILL.md.
1
Files scanned
343
Lines analyzed
0
Review items
0
False positives ignored
Confirmed security concerns (4)
High
Unconstrained MCP Server Command Execution
The skill asks agents to run placeholder server commands and package runners such as bunx and npx. Without validation or version pinning, a user-provided server command can execute untrusted code.
The file directly instructs MCP CLI execution using target-controlled SERVER_COMMAND and latest package runners. This is a legitimate workflow, but it executes external binaries.
High
Broad Home Directory Exposure in Example
The filesystem server example points at the user home directory. That can give an MCP server broad access to personal files if copied into a real workflow.
The example explicitly invokes a filesystem MCP server with the home directory as its scope. The risk depends on user execution, but the guidance is clear.
Medium
Broad Process Termination Pattern
The documented template repeatedly uses forceful process cleanup by pattern before running MCP shells. Broad pattern matching can kill unrelated processes or disrupt user work.
Multiple documented examples use forceful pattern-based cleanup. This can be safe for controlled test processes, but the template does not require exact process targeting.
Medium
Overbroad Skill Permission Templates
The current skill enables Bash, Read, Write, Edit, and Glob, and generated templates default to Bash, Read, and Write. This grants broad file and shell capabilities before target-specific need is established.
The frontmatter and generated templates explicitly grant broad tools. Some permissions are needed for this workflow, but least privilege is not enforced.
Risk Factors
⚙️ External commands (52)
SKILL.md:23 SKILL.md:32-42 SKILL.md:42-45 SKILL.md:45-51 SKILL.md:51-57 SKILL.md:57-63 SKILL.md:63-73 SKILL.md:73-77 SKILL.md:77-83 SKILL.md:83-98 SKILL.md:98 SKILL.md:100 SKILL.md:106 SKILL.md:110 SKILL.md:114 SKILL.md:116 SKILL.md:122 SKILL.md:123-135 SKILL.md:135 SKILL.md:137 SKILL.md:138-147 SKILL.md:147-154 SKILL.md:154-170 SKILL.md:170-171 SKILL.md:171-177 SKILL.md:177-190 SKILL.md:190-215 SKILL.md:215-217 SKILL.md:217-223 SKILL.md:223-227 SKILL.md:227-241 SKILL.md:241-243 SKILL.md:243-251 SKILL.md:251-257 SKILL.md:257-261 SKILL.md:261-265 SKILL.md:265-269 SKILL.md:269-273 SKILL.md:273-277 SKILL.md:277-282 SKILL.md:282-284 SKILL.md:284-287 SKILL.md:287-289 SKILL.md:289-292 SKILL.md:292-294 SKILL.md:294-297 SKILL.md:297-300 SKILL.md:300-330 SKILL.md:330-332 SKILL.md:332-337 SKILL.md:337-338 SKILL.md:338-341
🌐 Network access (3)
Audited by: codex View Audit History →
Share & cite this report
Share the versioned assessment report, neutral badge, embed card, and citations. Skillstore reports evidence without deciding whether this Skill is safe.
Copy report link
https://skillstore.io/skills/cygnusfear-create-mcp-skill/audits/8?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_reportMarkdown badge
[](https://skillstore.io/skills/cygnusfear-create-mcp-skill?utm_source=security_passport_badge)HTML badge
<a href="https://skillstore.io/skills/cygnusfear-create-mcp-skill?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/cygnusfear-create-mcp-skill/security.svg" alt="Skillstore security assessment" loading="lazy"></a>Embed card
<iframe src="https://skillstore.io/embed/skills/cygnusfear-create-mcp-skill.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>Academic citations (APA · BibTeX · CFF)
APA citation
Cygnusfear. (2026). create-mcp-skill security audit report (audit version 8) [Author version unspecified]. Skillstore. https://skillstore.io/skills/cygnusfear-create-mcp-skill/audits/8BibTeX citation
@techreport{cygnusfear-cygnusfear-create-mcp-skill-2026,
author = {Cygnusfear},
title = {create-mcp-skill security audit report (audit version 8)},
institution = {Skillstore},
year = {2026},
number = {8},
url = {https://skillstore.io/skills/cygnusfear-create-mcp-skill/audits/8},
note = {Author version unspecified}
}CITATION.cff
cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "create-mcp-skill security audit report (audit version 8)"
version: "unspecified"
type: report
authors:
- name: "Cygnusfear"
date-released: "2026-07-05"
url: "https://skillstore.io/skills/cygnusfear-create-mcp-skill/audits/8"
identifiers:
- type: other
value: "skillstore:cygnusfear-create-mcp-skill:audit:8"
description: "Skillstore immutable audit report identifier"
Skillstore Score
Why this score Evidence Confidence: Medium 55
Architecture
85
Maintainability
87
Content
72
Community
91
Spec Compliance
What You Can Build
Create a new MCP skill
Build a starter skill file after discovering a target MCP server and its tools.
Standardize MCP skill documentation
Apply a repeatable structure for setup, quick starts, tool notes, and troubleshooting.
Test MCP tool behavior
Exercise MCP tools with controlled inputs before adding examples to a published skill.
Try These Prompts
Create a basic MCP skill
Create a new skill for my MCP server. Ask for the server command, discover available tools, and draft the initial SKILL.md.
Document discovered tools
Review the MCP tools I discovered. Summarize the useful tools, required parameters, optional parameters, and safe example tasks.
Optimize command examples
Rewrite my MCP skill examples to use shell mode, timeouts, clean exits, and concise troubleshooting notes.
Audit an MCP skill template
Review this MCP skill draft for least-privilege tools, unsafe package execution, broad filesystem access, and unclear cleanup commands.
Best Practices
- Pin third-party MCP package versions before copying command examples.
- Test each MCP tool with harmless inputs before documenting workflows.
- Keep generated skill permissions limited to the required tools.
Avoid
- Using latest package versions in production skill examples.
- Running user-supplied server commands without confirmation.
- Copying broad process-kill patterns into unrelated skills.
Frequently Asked Questions
Does this skill run MCP servers automatically?
It provides command patterns, but execution still depends on the user's request and the agent's tool permissions.
What tools does it require?
It expects an MCP CLI plus a target MCP server package or binary.
Can it create a complete skill file?
Yes, it includes a starter SKILL.md structure and checklist, but server details still need editing.
Is it safe for untrusted MCP packages?
No, untrusted packages should be pinned, reviewed, and tested in a sandbox first.
Does it support Claude Code and Codex?
Yes, the workflow applies to Claude, Codex, and Claude Code when compatible tools are available.
Can I customize command examples?
Yes, replace placeholders with reviewed server commands, safe inputs, and exact cleanup targets.
Developer Details
Author
CygnusfearLicense
MIT
Skillstore revision
r1
Version notice
The author did not declare a version.
Ref
7e9e368dc1a370f2040369ebcd0db06d90cb48e5
Maintenance freshness
7/18/2026
Usage
5 downloads · 302 views
File structure
📄 SKILL.md