devops-infra MCP Server
Personal MCP server framework — deploy your AI backend to Vercel in 5 minutes. 86+ tools, 16 connectors. One endpoint for Claude Desktop, Cursor, Windsurf, ChatGPT.
Discovered via github-topic:model-context-protocol and last synced 1mo ago.
1. Install the package
npx @yassinello/create-kebab-mcp@latest
2. Add to claude_desktop_config.json
{
"mcpServers": {
"kebab-mcp": {
"command": "npx",
"args": [
"@yassinello/create-kebab-mcp@latest"
]
}
}
}Config file location: ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) / %APPDATA%\Claude\claude_desktop_config.json (Windows)
Extract structured data with AI (e.g., "get all prices from this page")
What it does
Get user profile: name, title, email, timezone, status
Read full page content as markdown
Fetch a LinkedIn company profile
Fetch a specific LinkedIn post
Hard-bypass fallback for stubborn paywalls
Get full issue details by identifier (e.g. ENG-123), including comments
List all accessible Airtable bases with IDs and permission levels
Get a single record by ID with all field values
Discover available actions for a specific app
List all webhooks that have received at least one payload
Restore skills and settings from a backup
`System/context.md`
—
Usually yes
List channels with topic, member count
Create a page in a database
Get company follower/employee insights
Full-text search across all issues
Create a new record with specified field values
Retrieve the last N payloads for a named webhook
No
Auth
Tools
Read recent messages from a channel
Update page properties and/or append content
Search Apify's actor marketplace
List projects with team filter, progress, and dates
Partially update a record (untouched fields are preserved)
Send a message or threaded reply
Query a database with filters and sorting
Run any Apify actor with custom input
Create an issue with name resolution for team, state, assignee, and labels
Case-insensitive text search on a specified field
View recent tool calls, errors, latency
`UTC`
**Requires:** `GITHUB_PAT` + `GITHUB_REPO` ### Browser Automation — 4 tools
Open URL, return visible text (handles JS-rendered pages)
Search messages (supports Slack operators: from:, in:, has:)
Read all replies in a thread
Fetch a LinkedIn person profile
Read paywalled articles via a reader service
Update an issue with same name resolution layer
Execute any action on a connected app (GitHub, Jira, HubSpot, Salesforce, Airtable, Linear, Figma...)
Clear internal caches (KV, API response, etc.)
Read LinkedIn feed (rate-limited, persistent session)
Search pages by title or content
Get recent posts from a LinkedIn company page
List issues with team, project, state, and assignee filters
List records with optional view, filter formula, sort, and limit
Retrieve the most recent payload for a named webhook
`30000`
Export skills and settings as a JSON backup
`User`
`en-US`
Execute actions: click, type, fill forms (natural language)
Get recent posts from a LinkedIn profile
List tables in a base with fields and views
Streaming transport diagnostic (verifies chunked transfer works end-to-end)
`main`
Auth
5+
DSN + token
Legacy
Protection
API token
Browserbase + OpenRouter
OAuth
Tools
`{ ok, version }` — add `?deep=1` for connector diagnostics
Optional HMAC
`MYMCP_DISPLAY_NAME`
PAT
API key
1 min
`MYMCP_CONTEXT_PATH`
`MYMCP_ERROR_WEBHOOK_URL`
`MYMCP_TIMEZONE`
Invoke any tool (playground API)
GitHub PAT
API key (1000+ apps via the bridge)
`MYMCP_LOCALE`
MCP Streamable HTTP
Admin token
`MYMCP_TOOL_TIMEOUT`
Scheduled health check + webhook alert