Skill-MCP is a powerful LLM-managed skills platform that allows users to create, edit, and execute skills programmatically across various MCP-compatible clients like Claude and Cursor, streamlining the development process without manual uploads.
claude install fkesheh/skill-mcpSkill-MCP is a powerful LLM-managed skills platform that allows users to create, edit, and execute skills programmatically across various MCP-compatible clients like Claude and Cursor, streamlining the development process without manual uploads.
[{"step":"Define the Skill Purpose","action":"Identify the specific task you want the Skill-MCP skill to automate (e.g., code review, CI/CD checks, documentation generation). Use the prompt template to draft the skill's JSON configuration.","tips":["Focus on repetitive tasks to maximize efficiency gains.","Include clear input/output requirements in the skill definition."]},{"step":"Create the Skill File","action":"Save the JSON configuration to a file named `[SKILL_NAME].json` in the `/.cursor/skills/` directory. Ensure the file follows the MCP Servers schema.","tips":["Use the Cursor MCP documentation to validate the JSON structure.","Test the skill locally before deploying to a team."]},{"step":"Register the Skill in Cursor","action":"Open Cursor, navigate to the Skills panel (Ctrl+Shift+P > 'Skills: Open Skills Panel'), and click 'Add Skill'. Select the JSON file you created.","tips":["Cursor will automatically detect and register the skill.","Verify the skill appears in the Skills list with a green status indicator."]},{"step":"Execute the Skill","action":"Run the skill using the Cursor command palette (Ctrl+Shift+P > 'Skills: Run Skill') or via the CLI with `cursor skill run [SKILL_NAME] --input [PARAMETERS]`.","tips":["Use `--help` to display available parameters for the skill.","Check the Cursor output panel for real-time logs and errors."]},{"step":"Monitor and Iterate","action":"Review the skill's output and logs. Refine the skill's configuration or script based on feedback and edge cases encountered.","tips":["Log detailed errors in your skill's script to aid debugging.","Share the skill with your team and gather input for improvements."]}]
Automating skill creation for AI agents
Integrating skills into existing workflows
Enhancing AI capabilities without manual intervention
Rapid prototyping of AI functionalities
claude install fkesheh/skill-mcpgit clone https://github.com/fkesheh/skill-mcpCopy the install command above and run it in your terminal.
Launch Claude Code, Cursor, or your preferred AI coding agent.
Use the prompt template or examples below to test the skill.
Adapt the skill to your specific use case and workflow.
Create a new Skill-MCP skill in Cursor for [SKILL_NAME] that performs [SPECIFIC_TASK]. The skill should include a [BRIEF_DESCRIPTION], [INPUT_REQUIREMENTS], and [OUTPUT_FORMAT]. Use the following structure for the skill file: ```json { "name": "[SKILL_NAME]", "description": "[BRIEF_DESCRIPTION]", "author": "[YOUR_NAME]", "version": "1.0.0", "mcpServers": { "[SKILL_NAME]": { "command": "[COMMAND_TO_RUN]", "args": ["--input", "${input}", "--output", "${output}"] } } } ``````json
{
"name": "code-review-automation",
"description": "Automates code review tasks by analyzing pull requests for common issues, security vulnerabilities, and style violations. Generates a detailed report with actionable feedback for developers.",
"author": "DevOps Team",
"version": "1.0.0",
"mcpServers": {
"code-review-automation": {
"command": "python",
"args": [
"${workspaceFolder}/scripts/code_review.py",
"--pr-url", "${input.pr_url}",
"--repo-path", "${input.repo_path}",
"--output", "${output.review_report.md}"
]
}
}
}
```
### Example Output:
**Skill File Created:**
- **Location:** `/.cursor/skills/code-review-automation.json`
- **Status:** Successfully registered in Cursor MCP.
**Execution Command:**
```bash
cursor skill run code-review-automation --pr-url "https://github.com/org/repo/pull/42" --repo-path "/Users/dev/repo"
```
**Generated Output (review_report.md):**
```markdown
# Code Review Report for PR #42
## Overview
- **Repository:** org/repo
- **Pull Request:** #42
- **Reviewer:** DevOps Team
- **Date:** 2023-11-15
## Issues Found
### Critical
1. **SQL Injection Risk** (File: `src/db/user_queries.py`, Line: 45)
- **Issue:** Direct string concatenation in SQL query.
- **Fix:** Use parameterized queries.
- **Severity:** High
2. **Memory Leak** (File: `src/services/cache_service.py`, Line: 120)
- **Issue:** Unclosed file handle in `load_cache()`.
- **Fix:** Add `finally` block to ensure cleanup.
- **Severity:** High
### Warnings
1. **Hardcoded Credentials** (File: `config/prod.env`, Line: 3)
- **Issue:** Database password exposed in plaintext.
- **Fix:** Use environment variables or secret management.
- **Severity:** Medium
2. **Missing Unit Tests** (File: `tests/test_api.py`)
- **Issue:** Only 60% test coverage for new endpoints.
- **Fix:** Add tests for `/api/v2/users` endpoint.
- **Severity:** Low
## Suggestions
- **Performance:** Optimize `get_user_data()` in `src/services/user_service.py` to reduce query time by 30%.
- **Security:** Rotate API keys after this PR is merged.
- **Documentation:** Update README with new endpoint usage examples.
## Next Steps
- [ ] Address critical issues before merging.
- [ ] Schedule a security review for hardcoded credentials.
- [ ] Add missing unit tests.
```
**Cursor MCP Logs:**
```
[2023-11-15 14:30:22] INFO: Skill 'code-review-automation' executed successfully.
[2023-11-15 14:30:25] INFO: Review report generated at '/Users/dev/repo/.cursor/skills/output/review_report.md'.
```Take a free 3-minute scan and get personalized AI skill recommendations.
Take free scan