Menu
Load when the user shares a URL, asks to read a webpage or document, search the web for current information, or process online files (HTML, PDF, CSV, JSON).
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Load when building, editing, or debugging Mako React apps — app files, npm dependencies, data bindings, the @mako/app-sdk hooks (useQuery / useDuckDB / useLocation / useSearchParams / navigate), URL state and shareable deep links, materialized Parquet/DuckDB bindings, and the live preview runtime.
Load for dbt branches, commits, pushes, pull requests, merges, repository sync, or branch cleanup.
Load when creating, editing, scheduling, running, or debugging dbt jobs and production deployments.
Load when building, editing, running, or debugging dbt models, dbt projects, schema.yml tests, sources, seeds, snapshots, incremental models, materializations, or dbt jobs in the Transforms section.
Load when creating, editing, configuring, validating, or debugging database sync flows, connector flows, pagination, incremental sync, destination mappings, and flow form fields.
Add a new MCP server preset (connector) to Mako's MCP client system — server URL, auth model (OAuth DCR, pre-registered app, or API key), scopes, icon, tests, and docs. Use when connecting a new external MCP server (like Slack or Close CRM) or changing how MCP connections authenticate.
| name | web |
| description | Load when the user shares a URL, asks to read a webpage or document, search the web for current information, or process online files (HTML, PDF, CSV, JSON). |
| entities | ["url","link","webpage","website","article","pdf","fetch","search","scrape","docs"] |
Use web tools to gather information from public URLs on the internet.
web_search — discover current information. Returns { title, url, snippet }[]. Use when you need to find pages, not when the user already pasted a URL.fetch_url — read a specific URL in full. Handles HTML, PDF, CSV, JSON, and plain text. Does not execute JavaScript.fetch_url directly.web_search, then fetch_url on the best 1–2 results.http/https URLs only — no authenticated pages, no internal networks.max_chars only when necessary.