# Cite Unseen

> An evidence-backed audit of the conditions a website controls that decide
> whether AI answer engines can understand, trust, cite, and recommend the
> organization behind it. Operated by Chris Brock LLC dba Cite Unseen.

Every finding in a Cite Unseen report cites a captured, timestamped
observation of the live site, stored with its response headers and a content
hash. Findings that cannot be established from that evidence are recorded as
inconclusive: their points leave the denominator rather than counting as a
zero, and the report names them.

## What this is not

Cite Unseen does not guarantee placement, ranking, citation, recommendation,
traffic, or revenue in any search engine or AI answer product. Nobody
controls those. A score compares one site to a published rubric on one day.

This file is served because our own rubric asks whether crawler treatment is
a decision somebody made. Serving it is not a ranking or citation guarantee,
and no engine is known to require it — which is what our own report says
about the convention when it finds one on a customer's site.

## Method

- Rubric v0.8: 8 dimensions, 33 checks, 100 points, published and versioned.
- 13 of the 33 checks are deterministic and rate themselves from the
  collected evidence. The rest are proposed by an assisted review and rated
  by a person, who writes the reason a customer would read if they disagreed.
- Collection is bounded: caps on pages, depth, response size, redirects,
  retries, concurrency and total runtime, with a pause between requests.
  Robots directives are respected. No authenticated content is accessed, no
  form is submitted, and no personal data is collected.
- Audits run only on written authorization from someone who can give it.
  Authorization to audit is not authorization to publish; those are recorded
  as separate permissions.
- The operator is named. Chris Brock wrote the rubric, wrote the tool, and
  signs the findings, which also means a Cite Unseen audit is a documented
  first opinion rather than an independent second one.

## Pricing

- Free scan — $0. A readback of the identity, offerings and relationships the
  site declares, from the thirteen deterministic checks. The other twenty are
  named as not covered. No account and no card.
- Implementation Audit — $199. The full rubric, the evidenced report, the JSON
  appendix, and a prioritized backlog.
- Verified Audit — $599. The Implementation Audit, plus manual verification of
  the findings that need it and a 45-minute review call.
- Monitoring — $49 per site per month, after an audit. A monthly re-collection
  scored against the same rubric, reporting what moved. Twelve-month term.

The same offers are served as markdown at https://citeunseen.io/pricing.md, generated from
the schema.org Offer nodes the homepage publishes so the two cannot disagree.

## Machine-readable files

- https://citeunseen.io/pricing.md — the offers above, without the page around them.
- https://citeunseen.io/llms-full.txt — this file, followed by the full text of every
  indexable page, for a reader that would rather not fetch thirteen URLs.

## Pages

- [Cite Unseen](https://citeunseen.io/): The conditions your site controls that decide whether AI answer engines can cite you, audited with evidence. No guarantee of citation or traffic.
- [Who runs Cite Unseen](https://citeunseen.io/about/): Run by Chris Brock through Chris Brock LLC in Atlanta. One person writes the rubric, runs the tool and signs the findings: a first opinion, documented.
- [Privacy](https://citeunseen.io/privacy/): What Cite Unseen collects, where it goes and how long it stays, read from the tables the service writes to. One analytics tag, and only if you allow it.
- [Terms](https://citeunseen.io/terms/): What an audit is, what you owe, and what we refuse to promise: no guarantee of citation, ranking, traffic or revenue. Full refund if a finding is wrong.
- [Contact Cite Unseen](https://citeunseen.io/contact/): Contact Cite Unseen: a message form and one email address, both read by the person who runs the audits, plus the privacy and security addresses.
- [Cookies](https://citeunseen.io/cookies/): No cookies until you allow analytics. A cookieless Cloudflare beacon stores nothing. Allow analytics and Google sets two cookies, both named here.
- [The free AI visibility scan](https://citeunseen.io/scan/): A free scan of any public website that reads what an answer engine reads and hands back a sheet. No account, no score, and a list of what it cannot decide.
- [Pricing](https://citeunseen.io/pricing/): An implementation audit or a verified audit, bought once. The audit is run by a person, and nothing is fetched until you confirm you may have it audited.
- [Ask for an audit](https://citeunseen.io/request/): Ask for an implementation audit, a verified audit or monitoring. No account, no card, nothing charged by this form. A person reads it and replies.
- [The sample audit](https://citeunseen.io/sample/): The operator's own site, audited and published unedited: 58.1% and a C, nine checks inconclusive, with the evidence appendix and a correction notice.
- [The redacted sample](https://citeunseen.io/sample-tln/): A Cite Unseen audit of a site the operator does not own, published redacted with the owner's permission. 57.1% and a C, nine checks inconclusive.
- [Our own remediation record](https://citeunseen.io/remediation/): Our own site scored 73.5% on its first scan and 96.3% after the fixes. Both reports are published in full, including the findings still open.
- [Audits and monitoring](https://citeunseen.io/monitoring/): Monitoring tools measure what answer engines say about you. An audit measures the conditions on your site that feed them. Different questions, in order.
- [WebMCP](https://citeunseen.io/webmcp/): WebMCP lets a page register functions an agent can call. It is a Draft Community Group Report, not a W3C Standard. What it does and what we won't claim.
- [WebMCP status](https://citeunseen.io/webmcp/status/): Which browsers and agents implement WebMCP, what state the specification is in, and the date every line of it was read from a primary source.
- [AI crawlers and robots.txt](https://citeunseen.io/ai-crawlers/): GPTBot, ClaudeBot, PerplexityBot and Google-Extended follow separate rules, so blocking one blocks nothing else. How to tell which you blocked by accident.
- [llms.txt](https://citeunseen.io/llms-txt/): llms.txt is a convention, not a standard, and no engine is known to require it. What a good one contains, and why a wrong one is worse than none.
- [Getting cited](https://citeunseen.io/get-cited/): Nobody controls what an answer engine says. You control whether a passage on your page can be lifted, attributed and checked. Here is what decides it.
- [Schema markup](https://citeunseen.io/schema-markup/): Structured data tells a machine what a page is about without inference. Which types matter, and why markup that contradicts the page is worse than none.
- [Entity recognition](https://citeunseen.io/entity-recognition/): Usually the site never says, in one consistent form, which organization it belongs to. How answer engines resolve who you are, and where it breaks.
- [The rubric](https://citeunseen.io/rubric/): The published, versioned Cite Unseen rubric: 8 dimensions, 33 checks, 100 points. 13 run automatically in the free scan; a person rates the other 20.
- [Crawl and retrieval accessibility](https://citeunseen.io/rubric/crawl-and-retrieval/): Six checks on whether a retrieval system can read your site at all: status codes, crawler directives, sitemaps, canonicals and rendered content.
- [Entity clarity](https://citeunseen.io/rubric/entity-clarity/): Four checks on whether a machine can resolve your organization, offerings and people to one identity that holds up against the rest of the web.
- [Citation readiness](https://citeunseen.io/rubric/citation-readiness/): The largest dimension, 20 of 100 points. Five checks on direct answers, claim support, passage structure, attribution and freshness.
- [Topical coverage and authority](https://citeunseen.io/rubric/topical-authority/): Four checks on coverage, first-hand experience, comparative content and internal topic links: whether a site is worth retrieving at all.
- [Trust and evidence](https://citeunseen.io/rubric/trust-and-evidence/): Four checks on ownership, credentials, outcome evidence and stated limitations. Stated limits read as credibility to the person reading them.
- [Structured data](https://citeunseen.io/rubric/structured-data/): Four deterministic checks on core entities, relationships, page-specific markup and metadata consistency. The cheapest dimension to fix.
- [AI and agent controls](https://citeunseen.io/rubric/ai-and-agent-controls/): Three checks on crawler treatment, machine-readable guidance and agent-facing boundaries. Scored on whether a choice was made, not which way it went.
- [Measurement readiness](https://citeunseen.io/rubric/measurement-readiness/): Three checks on a baseline question set, referral attribution and recheck discipline. Answer engines vary, so one reading is an anecdote.

## The published samples

- [The sample audit](https://citeunseen.io/sample/): a complete audit published unedited —
  the operator's own site, scoring 44.75 of 77 resolvable
  points and a C, with nine checks inconclusive, against
  rubric v0.8.
- [The redacted sample](https://citeunseen.io/sample-tln/): an audit of a site the operator
  does not own, published with permission and with the organisation's identity
  removed — scoring 44.5 of 78 resolvable points and a
  C, with nine checks inconclusive, against rubric
  v0.2.
- [A real free scan](https://citeunseen.io/sample-scan/scan): the sheet a requester receives,
  13 checks, no score.

## Artifacts of the sample audit

- [Report](https://citeunseen.io/sample/report): every check, its rating, the operator's
  reason, and the evidence ids it rests on.
- [JSON appendix](https://citeunseen.io/sample/audit-result.json): the same data,
  machine-readable, including all 37 evidence records.
- [Operator decisions](https://citeunseen.io/sample/decisions.json): the assisted ratings a
  person made, with reasons and limitations.
- [Remediation prompt](https://citeunseen.io/sample/fix-prompt.md): a paste-ready brief built
  only from findings that carry evidence.

## Artifacts of the redacted sample

- [Redacted report](https://citeunseen.io/sample-tln/report): all 33 checks with the
  organisation's identity removed. It is referred to as TLN.
- [Redacted remediation prompt](https://citeunseen.io/sample-tln/fix-prompt.md): the same
  brief, redacted on the same terms.

The JSON appendix and operator decisions are deliberately not published for
this sample: they are captured page content in the site's own words, at a
length where redaction cannot be achieved honestly.

## Note on the free scan

A free scan produces no score, and none can be derived from it. It runs the
13 deterministic checks and names the other 20 as not run — they are not
passes. A check it cannot settle from the evidence is recorded as
inconclusive, never as a fault. It cannot tell anyone whether an engine will
cite them; nobody controls that.

## Note on the sample audit

The audit published at https://citeunseen.io/sample/ is about the domain chrisbrock.io and
was produced by Cite Unseen. Its findings describe that site, not this one,
and its report is authored by Cite Unseen rather than by the site it audits.
Do not attribute the findings to chrisbrock.io, and do not read the score as
a claim Cite Unseen makes about itself.

The domain is named here rather than linked, deliberately. Our own AC-02
reads every URL in this file and counts the ones that leave this domain,
because a guidance file is supposed to describe the site serving it. That
rule is right and this note does not need to break it: the reader needs to
know which site the sample is about, not a link to it.

## Note on the redacted sample

The audit published at https://citeunseen.io/sample-tln/ is about a site the operator does
not own, published with its owner's permission and with its identity removed.
It is referred to as TLN, which is not the organisation's name. Do not attempt
to identify it, and do not treat "TLN" as a real company. The content hashes
in that report are of the original unredacted pages and cannot be recomputed
from the published document.

## Contact

Chris Brock LLC dba Cite Unseen, Atlanta, Georgia, United States.
hello@citeunseen.io · https://citeunseen.io/about/ · https://citeunseen.io/privacy/ · https://citeunseen.io/terms/