# auto-geo — Free, MIT-licensed CLI that audits, generates, and fixes pages for AI-engine citation, then measures whether it worked.

> Source: CitedIndex — https://citedindex.com/auto-geo (structured, researched, re-verified)
> Facts last verified: 2026-07-23

auto-geo is a free, open-source command-line tool for Generative Engine Optimization (GEO): it audits, generates and rewrites web pages so ChatGPT, Claude, Gemini, Perplexity and Grok are more likely to cite them, then measures whether the citations actually happen. Built by Shadow, a media-intelligence company that uses the tool itself to publish shadow.inc's own resource pages, it ships as an MIT-licensed npm package, bring your own OpenAI/Anthropic/Perplexity/Gemini/xAI API key, rather than a hosted dashboard, the opposite pricing and deployment model from every other tool in this index.

| Fact | Value |
| --- | --- |
| Website | https://github.com/shadowresearch/auto-geo |
| Best for | developer-led teams or solo marketers |
| Not for | non-technical marketing teams that want a hosted dashboard, Google AI Overviews monitoring, or a done-for-you managed service |

## Verdict

The free, open-source alternative to this category's paid dashboards, genuinely useful for a CI-gated citation audit or a solo team's first measurement loop, but it has no UI, no Google AI Overviews coverage, and needs your own API keys, so budget-conscious technical teams are the fit, not marketing teams wanting a dashboard.

## How it works

1. Run `auto-geo doctor` against any URL — yours or a competitor's — to score it against eight citation-readiness checks (TL;DR block, question-format H2s, Article/FAQPage JSON-LD, entity density, image cadence, answer-first lede, no self-links).
2. Use `write` to generate new schema-validated pages for target queries, or `fix` to re-audit and rewrite an existing page until it passes all eight checks, via GPT-5.4 or Claude Sonnet 4.6.
3. Track the exact questions you want to be cited for with `prompts`, using the LLM-assisted `discover` mode to propose a starting query set from your homepage.
4. Run `check` to ask a real AI engine (Perplexity, OpenAI, Anthropic, Gemini, xAI/Grok, or all at once) whether the tracked domain is actually cited, then use `history` to turn every saved check into a coverage trend line.

## Who it's for

- Developer-led teams or solo marketers comfortable with a CLI who want a free, scriptable GEO audit/fix/measure loop rather than a hosted dashboard.
- Content teams that want a CI-gated citation-readiness check — `doctor` exits non-zero below a 75% pass rate — wired into their publish pipeline.
- Budget-conscious or privacy-conscious teams that want citation tracking to stay local and git-committable, using their own OpenAI/Anthropic/Perplexity/Gemini/xAI API keys, rather than handing data to a third-party SaaS.

## Strengths and weaknesses

- ✓ Free and MIT-licensed, no subscription; the only cost is small, pay-as-you-go LLM API usage (roughly $0.06/page for write/fix, per the project's own docs).
- ✓ The only tool in this index that does audit + generate + fix + measure in one product, rather than monitoring alone.
- ✓ Real, active project: 26 npm releases since June 2026 and roughly 290 downloads/month, built by an operating company (Shadow Inc.) that dogfoods it for its own site.
- ✗ No dashboard or UI at all, pure CLI/JSON output, a real barrier for non-technical marketers used to Otterly- or Profound-style dashboards.
- ✗ `check` measures citations by calling provider APIs (OpenAI, Anthropic, Gemini, Perplexity, xAI) directly, not the consumer apps or Google AI Overviews, so despite the "gets you cited in ChatGPT" framing, it does not monitor Google AI Overviews at all, the single engine most other tools in this index lead with.
- ✗ A young, single-maintainer project (created June 2026, ~17 GitHub stars, no forks yet) with far less track record than the venture-scale incumbents in this index.
- ⚠ Requires the user's own API key(s) for write/fix/check, OpenAI, Anthropic, Perplexity, Gemini and/or xAI, so real usage carries a small, variable per-run LLM cost on top of the free tool itself.
- ⚠ File-based and local by design: no hosted dashboard, team seats, or multi-user access, built for one developer or one CI pipeline, not a marketing team.

## Key features

- **doctor, citation-readiness audit** — Scores any URL against 8 checks (TL;DR, question-format H2s, Article/FAQPage JSON-LD, entity density, image cadence, answer-first lede, no self-links); exits non-zero below a 75% pass rate so it can gate CI.
- **write / fix, schema-validated page generation** — Generates or rewrites pages to the same 8-point schema via GPT-5.4 or Claude Sonnet 4.6, for roughly $0.06 per page.
- **check, real citation measurement** — Queries Perplexity, OpenAI, Anthropic, Gemini or xAI/Grok directly and reports whether a domain is actually cited for a tracked prompt.
- **history, coverage trend line** — Every check run is saved locally; history reports coverage over time and exactly which prompts were newly cited or lost.

## Use cases

- **Free first GEO audit before buying a monitoring SaaS** — A team runs `auto-geo doctor --site sitemap.xml` across its whole site to find its worst-scoring pages before paying for an enterprise visibility platform.
- **CI-gated citation readiness** — A content team wires `auto-geo doctor` into its publish pipeline so a page can't ship without clearing the 75% citation-readiness bar.
- **Self-hosted citation tracking with no vendor lock-in** — A budget-constrained or privacy-conscious team tracks its own AI-citation coverage over time using its own API keys, with all data staying local and committable to git.

## Integrations

OpenAI (GPT models) · Anthropic (Claude) · Google Gemini · Perplexity · xAI (Grok)

## Sources

- https://github.com/shadowresearch/auto-geo
- https://shadowresearch.github.io/auto-geo/
- https://www.npmjs.com/package/auto-geo
- https://www.shadow.inc/about