← Back to research
·11 min read·company

GEO Optimizer Skill

GEO Optimizer (Auriti-Labs) is an open-source AEO/GEO toolkit — CLI, Python library, MCP server, and Astro integration — that audits sites against 47 research-backed methods and checks whether ChatGPT, Perplexity, and Gemini actually cite them. 463 stars, MIT license, 1,720 tests, solo maintainer.

https://geoready.dev ↗
agent-skillsseogeoopen-sourcepythonmcp

Key takeaways

  • The only tool in the SEO/GEO agent-skills category built as tested software rather than a prompt pack — a pip-installable CLI, Python library, MCP server, and Astro integration backed by 1,720 tests and an explicit academic foundation (Princeton GEO paper, KDD 2024; AutoGEO, ICLR 2026)
  • Closes the loop other audit skills leave open: `geo citations` queries live answer engines (BYO API key) to report whether your brand is mentioned and your domain cited — and which competitors get cited instead
  • Open-core economics — the CLI is MIT-licensed and free forever, while the GeoReady SaaS adds monitoring, score history, and agency reporting from $19/month; 463 stars as of June 11, 2026 with a near-daily release cadence but a single maintainer

FAQ

What is GEO Optimizer Skill?

An open-source Answer Engine Optimization (AEO) and Generative Engine Optimization (GEO) toolkit that scores a website 0-100 on AI-search readiness across 8 categories and 47 research-backed methods, generates fixes (robots.txt rules, llms.txt, JSON-LD schema), and checks whether ChatGPT, Perplexity, Gemini, and Google AI Overviews cite the site.

How much does GEO Optimizer cost?

The CLI, Python library, MCP server, and Astro integration are MIT-licensed and free. The hosted GeoReady platform offers a free no-account web audit, with Pro/Studio/Agency plans from $19/month for monitoring, score history, regression alerts, and agency reporting.

How does GEO Optimizer work with AI agents?

It is agent-agnostic: it ships an MCP server any MCP-capable agent can call, runs as a standalone CLI in CI/CD, embeds as a Python library, and integrates with Astro builds — unlike most category rivals, which are Claude Code skill bundles.

How is GEO Optimizer different from claude-seo?

claude-seo is a broad Claude Code skill covering all of SEO with GEO as one capability; GEO Optimizer is a narrower, GEO/AEO-only standalone program with deterministic scoring, 1,720 tests, and live citation checking against real answer engines.

Executive Summary

GEO Optimizer is an open-source Answer Engine Optimization (AEO) and Generative Engine Optimization (GEO) toolkit from Auriti-Labs: one command scores any website 0-100 on AI-search readiness, generates the exact fixes (robots.txt bot rules, llms.txt, JSON-LD schema, citable content patterns), and — the part most audit tools skip — queries live answer engines to report whether ChatGPT, Perplexity, Gemini, and Google AI Overviews actually mention your brand and cite your domain.[1][2] Its 47 scoring methods are explicitly anchored to academic work: the Princeton-led GEO paper (KDD 2024) and AutoGEO (ICLR 2026).[1][3][4]

Despite "skill" in the repository name, this is not a Claude Code prompt pack. It is a pip-installable Python program — 15 CLI commands, a library API, an MCP server, and an Astro integration — backed by 1,720 tests and CI, which makes it the most software-engineering-grade entry in the SEO/GEO agent skills category.[1][5][6] The trade-offs are scale and concentration: 463 stars as of June 11, 2026 (created February 18, 2026), essentially one maintainer — Juan Camilo Auriti of Auriti Labs, Italy, with 517 of the repo's human commits — and an open-core model funneling toward the GeoReady SaaS (from $19/month).[1][7][8]

AttributeValue
CreatorAuriti Labs (Juan Camilo Auriti), Italy[7][1]
CreatedFebruary 18, 2026; last pushed June 11, 2026[1]
GitHub Stars463 (54 forks) as of June 11, 2026[1]
LicenseMIT (CLI/library); proprietary GeoReady SaaS on top[1][8]
Latest releasev4.14.0 "Quiet Glass" (June 11, 2026)[9]
FundingNot publicly disclosed; donations + GeoReady subscriptions[1]

Product Overview

The core loop is geo audit --url https://yoursite.com: the tool fetches the site, scores it 0-100 across 8 categories — robots.txt AI-bot permissions (27 bots across training/search/user tiers), llms.txt presence and depth, JSON-LD schema richness, meta tags, content structure, brand/entity coherence, freshness signals, and AI-discovery files — and emits prioritized fixes, with score bands from Critical (0-35) to Excellent (86-100).[1] geo fix --apply auto-generates missing files; geo llms builds an llms.txt from a sitemap; geo schema generates JSON-LD.[1]

The differentiating commands are outcome-side. geo citations sends real customer-style questions to answer engines (bring your own API key; Perplexity recommended for real web citations) and reports whether your brand is mentioned, whether your domain is cited as a source, and which competitors get cited instead; geo snapshots, geo drift, geo history, and geo track archive answers, detect regressions, and produce HTML trend reports.[1][2] A separate Citability Score (0-100) applies the 47 methods, weighted per the research lineage — Quotation +41%, Statistics +33%, Fluency +29%, Cite Sources +27%, and 43 more.[1][3]

Key Capabilities

CapabilityDescription
Site audit0-100 score across 8 categories, 15 CLI commands, 7 output formats[1]
Citation checkingLive brand-mention and domain-citation queries against real answer engines[1]
Fix generationAuto-generates robots.txt rules, llms.txt, JSON-LD schema, meta tags[1]
MonitoringHistory, drift detection, regression alerts (--fail-on warning for CI), trend reports[1]
Bonus diagnosticsCDN crawler blocking, JS-rendering dependence, prompt-injection detection, RAG chunk readiness, content-decay prediction, per-platform citation profiles[1]

Product Surfaces

SurfaceDescriptionAvailability
CLIpip install geo-optimizer-skill, or zero-install via uvx; runs in CI/CDGA, MIT[5]
Python libraryProgrammatic audit/score APIGA, MIT[1]
MCP serverExposes audits to any MCP-capable agent (Claude, etc.)GA, MIT[1]
Astro integrationBuild-time GEO checks for Astro sitesGA, MIT[1]
GeoReady webFree single-URL audit, no account; llms.txt generatorFree[8]
GeoReady Pro/Studio/AgencyHosted monitoring, score history, regression alerts, agency reportingFrom $19/month[8]

Technical Architecture

GEO Optimizer is deterministic Python software rather than LLM-prompt orchestration: audits are rule- and parser-based against the 47-method rubric, so the same input yields the same score — which is what makes the CI/CD story (geo drift --fail-on warning) credible.[1] The repo runs CI with codecov tracking and advertises 1,720 passing tests.[1] Only the citation-checking commands call external AI APIs, with user-supplied keys.[1]

pip install geo-optimizer-skill
# or zero-install:
uvx --from geo-optimizer-skill geo audit --url https://yoursite.com

Release velocity is unusually high: v4.14.0 shipped June 11, 2026, one day after v4.13.0 — roughly four major-minor versions per month since the February 2026 creation date.[9][1]

Key Technical Details

AspectDetail
DeploymentLocal CLI / library / MCP server / Astro plugin; optional GeoReady cloud[1][8]
Model(s)None required for audits; BYO API keys (Perplexity recommended) for live citation checks[1]
IntegrationsMCP, Astro, CI/CD (JSON output, fail thresholds), PyPI distribution[1][5]
Open SourceMIT; 463 stars, 54 forks, 1,720 tests as of June 11, 2026[1]

Strengths

  • Real software, not a prompt pack — 1,720 tests, CI with coverage tracking, PyPI packaging, and deterministic scoring set it apart in a category dominated by markdown skill bundles.[1][5]
  • Closes the outcome loop — geo citations and answer snapshots measure whether AI engines actually cite you, not just whether your site is theoretically optimized; most category rivals stop at the audit.[1]
  • Explicit research grounding — the 47 methods and their weights trace to the Princeton GEO paper (KDD 2024) and AutoGEO (ICLR 2026), rather than folk SEO wisdom.[3][4][1]
  • Agent-agnostic surfaces — the MCP server, CLI, library, and Astro integration work with any stack, where most competitors require Claude Code specifically.[1][6]
  • Fast iteration — near-daily releases through early June 2026, with recent additions like RAG chunk readiness, content-decay prediction, and prompt-injection detection.[9][1]

Cautions

  • Bus factor of one — 517 of the repo's human commits come from a single maintainer (plus dependabot); there is no contributor community to absorb a maintainer departure.[1]
  • Modest adoption for the category — 463 stars as of June 11, 2026 versus 8,000-30,000+ for the category leaders; the project is four months old.[1][6]
  • GEO efficacy is contested ground — the tool inherits the open question of how much llms.txt and AI-discovery files actually move citations; its own README leans on vendor-ecosystem statistics for the "why" case, and the underlying research measures content edits, not site plumbing.[1][3]
  • Open-core funnel — the README doubles as marketing for the GeoReady SaaS (pricing links, signup CTAs), and monitoring continuity features are positioned as platform-only.[1][8]
  • No independent validation yet — no third-party benchmarks, case studies, or community review of the scoring rubric exist as of June 2026.[10]

What Developers Say

As of June 11, 2026, there is no substantive independent community discussion of GEO Optimizer: no Hacker News thread surfaces for the project on Algolia search, and no notable Reddit or independent review coverage was found.[10] That absence — for a four-month-old, 463-star project — is itself a data point: adoption so far appears search- and GitHub-driven rather than community-driven.[1] The only published commentary located is the maintainer's own writing (e.g., his Medium essay arguing GEO is becoming an operations problem), which is vendor voice and is not quoted here as community sentiment.[11]

Pricing & Licensing

TierPriceIncludes
CLI / library / MCP / AstroFree (MIT)All 15 commands, 8 scoring categories, citation checks (BYO API keys), CI/CD use[1]
GeoReady Free$0, no accountSingle-URL web audit, llms.txt generator, educational pages[8]
GeoReady Pro / Studio / AgencyFrom $19/monthHosted monitoring, score history, regression alerts, agency reporting[8]

Licensing model: Open core — the audit engine is MIT-licensed on GitHub and PyPI; the GeoReady platform layers proprietary server-side continuity (monitoring, history, teams) on top.[1][8]

Hidden costs: Live citation checking requires your own answer-engine API keys (Perplexity recommended), billed by those providers; ongoing multi-site monitoring effectively requires the paid platform.[1][8]

Competitive Positioning

Direct Competitors

CompetitorDifferentiation
claude-seoCategory leader on adoption with the broadest SEO surface (25 sub-skills, real search-data extensions); Claude Code-native prompts vs. GEO Optimizer's standalone tested CLI and narrower GEO/AEO focus[6]
MarketingSkillsThe category's star-count leader, covering all of marketing with SEO/GEO as one capability; GEO Optimizer is a single-purpose instrument by comparison[6]
geo-seo-claudeThe other GEO-first entry, but delivered as Claude Code subagents with PDF deliverables; GEO Optimizer counters with deterministic scoring, CI/CD integration, and live citation checks[6]
SEO/GEO Optimizer (199 Biotechnologies)Similar name, different project — a smaller, less actively maintained Claude skill[6]

When to Choose GEO Optimizer Over Alternatives

  • Choose GEO Optimizer when: you want GEO checks in CI/CD with deterministic, testable scoring; you need an MCP server or library rather than a Claude Code skill; or you want to measure actual AI citations, not just audit readiness.
  • Choose claude-seo when: you live in Claude Code and want the broadest SEO coverage with real search data.
  • Choose MarketingSkills when: SEO/GEO is one need among many across a marketing workload.

Ideal Customer Profile

Best fit:

  • Engineering teams that want AI-visibility checks as a CI gate alongside lint and tests
  • Developers building agents that need GEO auditing as a callable tool (via the MCP server or Python API)
  • Astro-based sites wanting build-time GEO validation
  • Agencies needing reproducible, scriptable audits across client portfolios (paired with GeoReady for reporting)

Poor fit:

  • Marketers who want a conversational, Claude Code-native workflow rather than a CLI
  • Teams needing broad traditional SEO (keywords, backlinks, local) — this tool is deliberately GEO/AEO-only
  • Buyers requiring a vendor with funding, a team, and SLAs behind the software

Viability Assessment

FactorAssessment
Financial HealthUndisclosed — solo open-source project monetized via GeoReady subscriptions (from $19/month) and donations[8][1]
Market PositionNiche challenger — 463 stars in a category whose leaders hold thousands; differentiated as the only tested-software, agent-agnostic entry[1][6]
Innovation PaceVery high — v4.14.0 within four months of creation, near-daily releases in June 2026[9]
Community/EcosystemThin — one human committer, 2 open issues, no HN/Reddit footprint as of June 2026[1][10]
Long-term OutlookHinges on one maintainer's persistence and on GEO/AEO consolidating as a real budget line[1]

The engineering signal (1,720 tests, CI, deterministic scoring, MCP/Astro/library surfaces) is far stronger than the adoption signal (463 stars, no community discussion), an inversion of the usual pattern in this category, where prompt packs with thin engineering hold thousands of stars.[1][6] The MIT license caps downside — the audit engine survives even if the maintainer or the GeoReady business does not.[1]

Bottom Line

GEO Optimizer is the engineer's entry in the SEO/GEO agent-skills category: a research-grounded, heavily tested, MIT-licensed audit engine that uniquely measures real AI citations and slots into CI/CD, MCP agents, and Astro builds — built and maintained by one person in Italy with no community footprint yet.

Recommended for: teams that want deterministic GEO checks in pipelines or agents, and developers who prefer auditable Python software over prompt bundles.

Not recommended for: anyone needing full-spectrum SEO, a Claude Code-native conversational workflow, or a vendor with organizational depth behind it.

Outlook: Watch whether the star curve bends as GEO budgets form, whether any independent validation of the 47-method rubric appears, and whether GeoReady revenue is enough to keep a near-daily release cadence sustainable for a solo maintainer.

Research by Ry Walker Research • methodology

Sources

Share on XShare on LinkedIn