geo-visibility-looplisted

# GEO Visibility Loop Generative Engine Optimization (GEO) is the practice of making a brand citable and recommendable by AI systems. This skill measures that visibility honestly and turns the measurement into a weekly improvement loop. ## Hard rules (apply in every mode) 1. **No invented numbers.** Every score is a judgment on measured findings and is labeled as such. Every count (mentions, stars, clicks) comes from a tool call or is reported as a data gap. 2. **A probe that was not run is a data gap, not a MISS.** Say which probes ran and which did not, and why. 3. **All-positive results are a warning sign.** If every category scores high, re-check the brand-authority probes; most sites fail off-site, not on-site. ## Configuration Read `geo-loop.config.json` from the project root (copy from `config.example.json` in this skill's directory). It defines: - `domain`: the site to audit and probe - `repos`: GitHub repos (owner/name) whose visibility should be tracked, may be empty - `queries.brand`, `queries.money`, `queries.repos`: the fixed probe set. Keep it stable between runs; a changed query invalidates the week-over-week diff for that line. - `state_file`: where the loop stores last week's results (JSON) - `report_dir`: where weekly reports are written ## Mode 1: `audit` (baseline, run once per site) 1. **Technical checks** (direct HTTP, no crawler needed): - `robots.txt`: are AI crawlers (GPTBot, ClaudeBot, PerplexityBot, Google-Extended) all