open-geo releasesRelease notes for the open-geo GEO visibility trackerhttps://pupok462.github.io/open-geo/2026-08-18T00:00:00Zopen-geo contributors0.4.1 — portable run artifacts and composable agent workflowstag:github.com,2026:open-geo/0.4.12026-08-18T00:00:00ZEvery run now exports open-geo.run-artifact.v1 for agent-to-agent handoff. Data mode starts no servers, the installed skill bootstraps its runtime on first use, and AgentsMesh generates adapters for Claude Code, Codex CLI, Cursor and Gemini CLI.0.4.0 — the PDF report stops being a subset of the dashboardtag:github.com,2026:open-geo/0.4.02026-08-18T00:00:00ZThe report gains a per-query results table grouped by outcome, a standalone gap list, a metric glossary, and dashboard parity in the existing sections. It also stops disagreeing with the dashboard: "whole period" now folds the period with the same weighted math as the API instead of printing the latest run under that label. The audit section finally prints the remediation it had been promising, and the layout was rewritten — no clipped text, no orphaned headers, page numbers.0.3.4 — scripted fast path for source extraction, live audit of all seven enginestag:github.com,2026:open-geo/0.3.42026-08-15T00:00:00ZEvery engine playbook was re-probed against the live interface, and two prescribed procedures turned out to work around a tool limit rather than the page. The result is a scripted fast path for source extraction — cheaper, and explicitly a fast path rather than a trusted one: the agent still cross-checks before the numbers count.0.3.3 — Perplexity engine, audit-gate robustnesstag:github.com,2026:open-geo/0.3.32026-08-12T00:00:00ZPerplexity joins the supported engines — seven, all live-validated — with its playbook rewritten against the real interface. Plus an audit gate that degrades instead of crashing on a malformed domain, a robots.txt 429 that no longer reads as "everything allowed", and a dashboard PDF endpoint that cleans up its temp files.0.3.2 — correctness and stability passtag:github.com,2026:open-geo/0.3.22026-08-08T00:00:00ZTen fixes to cases that could stop a run that should have proceeded, or report a number that looked right and was not: engine-aware audit gating, unreadable robots.txt, framework false positives, brand-name splitting, 1-based rank validation, and a leaderboard that now ranks server-side.0.3.1 — README rework, project page, structured datatag:github.com,2026:open-geo/0.3.12026-08-03T00:00:00ZA README that opens with a direct definition in four languages, a comparison section and expanded FAQ, and a GitHub Pages project page carrying SoftwareApplication and FAQPage structured data.0.3.0 — repeat runs, weekly trend, combined multi-engine PDFtag:github.com,2026:open-geo/0.3.02026-07-25T00:00:00ZRepeat runs with a min-max spread instead of a delta, an ISO-week rollup on the trend chart, a combined multi-engine PDF, and per-lens KPI cards.0.2.0 — brand mention rate and engine-neutral labelstag:github.com,2026:open-geo/0.2.02026-07-24T00:00:00ZA seventh metric for the share of answers naming the brand, a results-table gap filter, and engine-neutral metric labels.0.1.6 — GEO audit gate, DeepSeek and Perplexity playbookstag:github.com,2026:open-geo/0.1.62026-07-10T00:00:00ZA deterministic pre-run check of whether an AI engine can read the target domain, plus capture playbooks for DeepSeek and Perplexity.0.1.0 — first public releasetag:github.com,2026:open-geo/0.1.02026-07-03T00:00:00ZThe /open-geo command, capture playbooks for five engines, seven metrics over a nested funnel, question harvesting, and a four-language dashboard and PDF.