Opens in a new tabSkip to content
Agent LighthouseAgent Lighthouse

    Searches the text of every published page. The evidence sources themselves are not in this index — search all of them on the trusted sources page.

    GitHub ↗
    Browse checks and page contents
    operability-safety/trust-txt-reciprocity-coherence

    trust.txt associations are reciprocated and agree with robots.txt

    What it checks

    ROADMAP / UNSCORED. Parses trust.txt for publishers who maintain one, validating attribute names, resolving the reciprocal member=/belongto= relationships across domains, and cross-checking the datatrainingallowed= declaration against what robots.txt actually tells AI crawlers. Reported as an informational trust signal, never scored.

    Why it matters

    No search engine, platform or AI company documents reading trust.txt. The JournalList reference document (verified 2026-10-18) pairs belongto= on a member with member= on the association in its examples, without stating a reciprocity rule, and names no consumer. The JournalList FAQ (verified 2026-10-18) names only generic audiences: “the search engines, social networks and advertisers”. The one documented reader is a JournalList browser extension (verified 2026-10-18) that shows a badge to human readers. So the reciprocity check has no machine reader. The file still serves the JournalList network. For AI training preferences, robots.txt groups for the AI crawlers are the channel crawlers read.

    Evidence

    • Reference Document for trust.txt Specifications — JournalList.net (spec, verified 2026-10-18). Released 2020-06-29, revised through 2024-07-11; datatrainingallowed= added 2024-04-04. belongto= pairs with member= on the association, controlledby= with control=. Names no consumer.
    • Frequently Asked Questions — JournalList.net (verified 2026-10-18). Says “crawlers may look at the trust.txt file at any time” and names no reader.
    • trust.txt launches browser extension — Reynolds Journalism Institute, 2025-02-21 (verified 2026-10-18). A browser badge for human readers; “about 3,000 participating publishers”. No machine consumer is named.

    Grade: C, the policy’s community-convention row. The file is registered in the IANA Well-Known URIs registry (provisional; verified 2026-10-18) and published, and nothing documents a machine consuming it.

    How it scores

    Informative, weight 0. Every result carries an unconsumed notice. The audit never warns or fails. No trust.txt is not applicable. A trust.txt passes; each unreciprocated association, unknown attribute or datatrainingallowed= conflict with robots.txt is counted in the message and listed in details.observations, with the note that it does not affect AI agents.

    Example finding

    A regional news site publishes trust.txt with datatrainingallowed=no and belongto=example-press-association.org. That association’s trust.txt has no member= line for the site, so the membership claim is uncorroborated; meanwhile robots.txt has no GPTBot or ClaudeBot rules at all, so the site’s only enforceable channel grants exactly the training access its declaration refuses.

    Sources