Content-Signal directives are malformed

What it detects

robots.txt declares Content-Signal directives, but one or more are malformed — an unrecognized value (only yes/no are valid) or a token without the name=value form.

Why it matters

A declared-but-broken signal is worse than an absent one: it reads as an explicit content-use preference while giving intermediaries and agents nothing they can actually apply, so the site is treated as undeclared anyway.

How to fix

Fix the malformed Content-Signal directive(s) in robots.txt. Each dimension takes only `yes` or `no`, in `name=value` form, comma-separated: Content-Signal: ai-train=no, search=yes, ai-input=yes Valid dimensions: `search` (search indexing), `ai-input` (retrieval/grounding at answer time), `ai-train` (model training). An unrecognized value or a token without `=` is ignored by consumers, leaving the site effectively undeclared. See contentsignals.org for the grammar.

Example

Content-Signal: ai-train=maybe, search=yes # 'maybe' is not a valid value

References

• Content Signals specification
• RFC 9309 — Robots Exclusion Protocol
• Sitemaps protocol