← ClaudeAtlas

umami-seolisted

Query a self-hosted Umami instance and turn the results into evidence-based SEO recommendations. Use this skill whenever a task involves organic search traffic, landing pages, high-bounce pages, page titles, traffic sources or marketing channels, Core Web Vitals, or questions such as "why did my traffic drop", "which pages are performing", "where does my traffic come from", "audit my SEO", "improve my SEO", "what should I write about", or "look at my analytics". Also use it to simply retrieve raw traffic data from Umami — totals, time series, per-page metrics, exports — when the numbers are wanted without interpretation. Provides the umami-q CLI, which queries the Umami API directly, so recommendations rest on real numbers instead of generic advice.
toorop/umami-seo · ★ 0 · API & Backend · score 70
Install: claude install-skill toorop/umami-seo
# Umami SEO analysis Turn self-hosted Umami analytics into concrete SEO recommendations. The whole point of this skill is to stop SEO advice from being generic. Every claim you make should trace back to a number in the collected data, and you must know the limits of that data well enough not to over-claim from it. ## Execution modes — read this first This skill works in two very different environments. Determine which one you are in before doing anything else. **Mode A — you have a shell with network access** (Claude Code, a local agent, a terminal). The bundled CLI works fully. This is the intended mode. **Mode B — sandboxed code execution with no network** (Claude.ai skills, many hosted agent sandboxes). Anthropic's code execution sandbox has **no outbound network access at all**, so the CLI cannot reach the Umami instance — not even over a VPN, because there is no network to route. In this mode: - do **not** waste turns retrying the CLI; - if an Umami MCP connector is available in the session, use its tools instead (`get_website_stats`, `get_website_traffic`, `get_website_metrics`, `get_performance`, ...); - otherwise ask the user to supply data files — a `report` produced elsewhere, or exported CSV/JSON — and analyze those files locally, which the sandbox permits. Detect the mode cheaply with `umami-q check` (see below). If it fails with a connection error, you are in Mode B. ## Setup The CLI lives at `scripts/umami-q.sh`, relative to this file. Run it fro