今年在 Hacker News 推出的 7 款開源 AI 搜尋能見度追蹤工具，各自有甚麼用途？

The AI visibility space has a paid side and an open-source side. The paid side is crowded: Otterly ($29/mo), Peec AI ($95/mo), Scrunch ($250/mo), Profound (enterprise), AthenaHQ ($295/mo), Semrush AI Visibility ($99/mo), and a dozen more. The open-source side is smaller but growing fast. Seven projects launched on Hacker News in 2026, each taking a different angle on the same problem: can you check your AI search visibility without paying a subscription?

Here is what each one does, what it costs to run, and what it leaves out.

1. Elmo

Elmo is a self-hosted AEO platform, MIT licensed. It tracks how your brand appears across ChatGPT, Claude, Gemini, Grok, Perplexity, Copilot, DeepSeek, Mistral, and Google's AI Mode and AI Overviews. You supply your own LLM API keys, so the cost is whatever the API providers charge per query. No per-prompt subscription.

What it does: brand mention tracking, citation tracking, competitor benchmarking, sentiment. Coverage is unusually wide for open source (9+ engines).

What it leaves out: site audit. Elmo tracks whether you appear in AI answers. It does not check whether your site is crawlable or parseable. You need a separate tool for that.

Cost to run: API costs only. If you run 50 prompts across 9 engines weekly, expect $5-15/week in API fees depending on which models you use.

2. Canonry

Canonry is a CLI that tracks how ChatGPT, Gemini, and Claude cite your site. Self-hosted on SQLite, config-as-code in YAML, CLI and API are first-class.

What it does: visibility scores, citation-readiness checks, brand accuracy across providers over time. Cross-references with Google Search Console and Google Analytics for a combined view.

What it leaves out: no site audit, no fix recommendations. It tells you what is happening, not what to do about it.

Cost to run: API costs for the LLM queries. SQLite is free.

3. Auto-geo

Auto-geo is a file-based GEO CLI from Shadow Inc, an AI PR lab. npm install or brew install.

What it does: prompt tracking (prompts discover reads your homepage and proposes queries), citation checking across ChatGPT, Claude, Gemini, Perplexity, and Grok, history diffs (which prompts you won or lost since last run), and a doctor command that audits pages for AI-citation readiness.

What it leaves out: the doctor audit is lighter than a full site audit. It checks structure and "chunkability" but does not check 27 AI crawlers or run a full scoring rubric.

Cost to run: API costs for the LLM queries. npm package is free.

4. Lettertrace

Lettertrace is a BYOK (bring your own key) library for tracking AI search, MIT licensed. It is a library, not a full platform. You embed it in your own codebase.

What it does: tracks how your brand appears in AI answers using your own LLM harness. Designed for developers who want to build visibility tracking into existing systems.

What it leaves out: no UI, no dashboard. It is a library. You build the interface.

Cost to run: API costs only.

5. CiteLadder

CiteLadder is an open-source AI search visibility tracker, recently launched on HN.

What it does: tracks whether your site gets cited by AI search engines. Details are sparse since the project is new.

What it leaves out: too early to assess. The repo is fresh.

Cost to run: not specified.

6. GetCito

GetCito is an open-source AEO tool, MIT licensed, run by a marketing agency.

What it does: AI Crawlability Clinic that checks how well AI bots can reach and parse your content.

What it leaves out: maintenance. The public repo has had a single substantive code commit since it launched in late 2025. It is run by an agency, not a dedicated product team. Do not expect ongoing fixes or new engine support.

Cost to run: free, but you get what you pay for in terms of maintenance.

7. Ansvisor

Ansvisor is an open-source, cloud-ready AI visibility platform. It tracks mentions, citations, competitors, and Share of Voice across ChatGPT, Gemini, Perplexity, Claude, Copilot, Google AI Overviews, and Google AI Mode.

What it does: separates owned citations, competitor citations, and third-party sources. Cross-platform monitoring with historical trends.

What it leaves out: no site audit. Like Elmo, it tracks visibility, not crawlability.

Cost to run: open source for self-hosting. Managed cloud option listed as coming soon.

Where geo-optimizer fits

GEO Optimizer is the only open-source tool in this list that does a deterministic site audit rather than citation tracking. The 16 CLI commands, 47 content checks, 27 AI crawlers, and 9 scoring categories check whether your page is reachable, parseable, and quotable. It does not track whether you appear in AI answers over time. That is what the monitoring layer on geoready.dev is for.

The distinction matters. Citation trackers (Elmo, Canonry, Ansvisor, Auto-geo, Lettertrace, CiteLadder) tell you whether you appear in AI answers. Site auditors (geo-optimizer, GetCito) tell you whether AI crawlers can read your page in the first place. You need both: audit to fix the foundation, monitor to track whether the fixes worked.

Comparison table

What none of them have done

None of these tools, including mine, has published a calibration study testing whether their scores or metrics predict actual citations. I did, and the result was null. That study is at github.com/Auriti-Labs/geo-optimizer-skill/tree/main/research/calibration-2026-10, with pre-registration hash and raw data.

If you use any of these tools, ask the maintainers to do the same. A score that has not been tested against reality is a number, not a metric.

If you want to run a free site audit, it is on geoready.dev. The CLI is on GitHub, 1,137 stars, MIT licensed, 2,000+ tests.