Content-Signal declaration does not match the documented syntax
content_signals_directive_invalid · Issue
Correct the known key/value to the syntax shown by its source. Use yes or no for search, ai-input and ai-train, or immediate, reference or full for the documented use extension. Choose the value that expresses your policy; this check does not choose it for you.
Severity: Notice. This is a concrete declaration defect in a published subset, not a claim that all consumers ignore the line.
Why it matters
A declared usage preference should be understandable in the syntax you intended. Consumer adoption and fallback behavior are not established by this audit.
What to do
- Review the original line, documented key, entry, line number and byte position.
- Use yes/no for search, ai-input and ai-train, or immediate/reference/full for the documented use extension.
- Correct the source template and run a full audit of the same site.
When to leave it as it is: An unknown extension or another unverified syntax is not a finding under this conservative rule. Check its publisher specification separately.
What we read
The original complete, directly fetched, same-origin robots.txt captured with HTTP 200, and its shared lexical index.
Exact published root-level entries for search/ai-input/ai-train=yes|no and use=immediate|reference|full. The published managed use=reference control is clean.
Example & evidence
Illustrative same-site declaration correction; no live result.
User-agent: *
Content-Signal: ai-train=maybe
Allow: /User-agent: *
Content-Signal: ai-train=no
Allow: /Confirm & check again
Check for yourself: Compare the actual served directive with the publisher’s documented form, including case, separators and extensions.
Check again: Full audit — A subsequent full audit of the same site rechecks the declaration with the same evaluator and lifecycle. Unknown or insufficient coverage never proves a fix.
Limits & sources
Unknown extensions, case/spacing/separators, path-scoped lines, an entirely empty Content-Signal directive and group ambiguity remain not checked. Known bad entries survive separately delimited unknown companions; those unknowns block reconciliation.
Repeated effective declaration groups and damaged raw fields, including known Unicode look-alikes, cannot look like checked removal.
A fully checked correction or complete declaration removal can reconcile disappearance. Partial, redirected, non-UTF-8, unreadable or oversized files cannot.
Template comments identify at most a template match, not Cloudflare management or hosting. The expired individual IETF draft is vocabulary context, not a grammar standard.
Confidence: High High for the exact original documented-key defect. Group uncertainty and other unverified entries remain independent coverage unknown.
Sources
- Cloudflare, Content Signals Policy — Documents the named usage signals and yes/no examples. Checked 4 October 2026.
- Cloudflare, Managed robots.txt — Documents the managed template and its usage extension; no hosting attribution is inferred. Checked 4 October 2026.
- Content Signals, Content Signals — Provides the published generator and documented use values. Checked 4 October 2026.
Editorial review: 2 November 2026