Different Search Surfaces, Different Evidence
OpenSEO covers more than keyword and competitor research, but the same evidence-and-control model does not produce the same kind of result everywhere. This chapter compares site audits, rank tracking, local search, link prospecting, and AI visibility so the reader can choose the right surface for a question and interpret blocked, missing, partial, cached, and genuinely negative results without collapsing them into one score.
The distinction is practical. A national organic result is not a Maps result. A crawl that cannot read a page is not proof that the page is broken. A backlink prospect is not a confirmed contact. An absent AI mention is not automatically evidence of absent brand awareness. Each surface has its own inputs, lifecycle, provider limits, and threshold for a responsible conclusion.
An audit is a background evidence pipeline, not an instant page score
OpenSEO's site audit starts a crawl that runs in the background. It is robots-aware and same-origin, and it checks crawled pages for issues such as broken links, missing or duplicated titles and descriptions, redirect chains, orphan pages, canonical problems, thin content, and other technical conditions. The agent does not receive a final answer at the moment it starts the job. It receives an audit identifier, waits for status, and then reads prioritized issues or per-page results.
That lifecycle changes how the result should be used. A completed audit can show what the crawler observed and attach concrete remediation guidance to an issue. A failed audit may still retain partial results for the pages it reached. A page that was blocked or rate-limited is a different state from a page that was fetched and found to be non-indexable. The workflow should preserve that difference instead of flattening all unreachable pages into “SEO errors.”
Lighthouse is optional for agent calls and is off by default because it adds substantial wall-clock time. That choice illustrates a useful design principle: the most detailed measurement is not automatically the best first measurement. If the immediate question is whether important pages are indexable, whether titles are missing, or whether a site has broken links, adding performance analysis may delay the decision without improving it.
The audit skill narrows the final report further. It asks for one action a site owner can take this week, verifies important findings against the live page, and avoids overwhelming a beginner with every detected issue. This is not a claim that only one issue exists. It is a decision rule for turning a broad crawl into an accountable next step. The evidence is the crawl; the recommendation is the human-facing interpretation; the one action is a prioritization choice.
Search Console and Analytics can strengthen that interpretation. A technical finding matters differently on a page that already earns impressions and conversions than on an abandoned test page. First-party data can tell the agent what the site is actually receiving, while the audit tells it what the crawler can observe. Neither source alone explains the entire business consequence.