generated: '2026-09-12'
method: searched
source: >-
  https://www.journeys.com/llms.txt and https://www.journeys.com/llms-full.txt
  (provider-published), plus live probes of www.journeys.com, fetched 2026-09-12.
applies_to: >-
  The ONLY machine-addressable surface Genesco publishes: the Journeys verified-bot search
  URL and the server-rendered JSON-LD on Journeys product detail pages. There is no API, so
  the usual API conventions (versioning, pagination envelopes, error envelopes, idempotency
  keys) have no subject here. Each is recorded below as the provider actually leaves it,
  rather than invented.

auth_style:
  scheme: none
  detail: >-
    "No API key, bearer token, custom request header, or separate API hostname is required
    for eligible bots." Access is decided at the edge by verified-bot identification, which
    the caller cannot obtain by presenting a credential.
  source: https://www.journeys.com/llms.txt

transport:
  protocol: HTTPS
  methods: [GET, HEAD]
  content_type_returned: text/html (condensed semantic HTML for eligible bots; full storefront otherwise)
  javascript_required: >-
    No for /search when the caller is an eligible verified bot, and no for /product and
    /kidz/product pages, which are server-side rendered. Yes for product listing pages and
    all /help articles, which are client-side rendered.

identification_and_attribution:
  user_agent: >-
    "Automated clients must identify themselves accurately" — the robots.txt named-agent
    group is the allow list (OAI-SearchBot, ChatGPT-User, Claude-SearchBot, Claude-User,
    PerplexityBot, Perplexity-User, Google-Agent).
  query_parameter: llm
  rule: >-
    Append llm={bot-name} to every Journeys URL an assistant links to — "?llm=" when the URL
    has no query string, "&llm=" when it does. Lowercase enumerated values; use "other" when
    the assistant cannot confidently identify its platform.
  values: [chatgpt, claude, perplexity, gemini, microsoftcopilot, meta, deepseek, other]

pagination:
  style: none-documented
  detail: >-
    The search response is described as returning a reported result count and a displayed
    product count, but no page/offset/cursor parameter is documented. Treat the response as
    a single page of discovery data.

versioning:
  style: none
  detail: No version segment, header, or date-based version — the surface is a website path.

error_semantics:
  style: http-status-only
  detail: >-
    No error envelope, no RFC 9457 problem+json, no error code registry. Failures surface as
    ordinary HTTP statuses and, for unverified automated clients, as a Cloudflare bot-management
    interstitial. Observed 2026-09-12: GET /search?keywords=black%20converse&llm=claude returned
    403 with a Cloudflare "Attention Required!" HTML body under both a browser User-Agent and a
    Claude-User User-Agent.

rate_limit_signaling:
  headers: none-observed
  detail: >-
    No X-RateLimit-*, RateLimit-* or Retry-After headers are documented or were observed. The
    provider states only "respect applicable crawl limits and site policies" with no published
    numbers. See rate-limits/genesco-rate-limits.yml.

caching:
  detail: >-
    "Search results may be cached and should be treated as discovery data. Product Detail Pages
    remain the authoritative source for product-specific facts."

data_freshness:
  rule: >-
    Do not guess live availability, prices, promotions, coupon eligibility, shipping dates, size
    availability, or store inventory; confirm from the Product Detail Page JSON-LD or send the
    user to the page.

idempotency:
  coverage: na
  mechanism: none
  scope: []
  detail: >-
    NOT APPLICABLE — read-only surface. Only GET and HEAD are supported and no mutating
    operation is published, so there is nothing to replay-protect. This is an honest `na`, not a
    missing mechanism: a read-only provider cannot double-fire a write.

dry_run_mode:
  status: na
  detail: No write surface, therefore nothing to rehearse.

reversibility:
  status: na
  detail: >-
    NOT APPLICABLE — Genesco publishes no machine-callable write operation of any kind. There is
    no create, update, delete, order, cart, or payment action an agent can take through a
    documented surface, so there is no action to reverse and no reversal window to state. If
    Genesco ever publishes an agentic checkout or order surface, this block becomes a real
    requirement and must record the cancel/return operation and its window from the provider's
    own returns policy.
  write_surfaces: []
  related_consumer_policy:
    note: >-
      For completeness and NOT as an agent-reversibility claim: the Journeys storefront's own
      Organization JSON-LD carries a hasMerchantReturnPolicy node, and the human returns policy
      lives at https://www.journeys.com/customer-service/returns-and-exchanges. That is a
      consumer return path a person exercises in a browser — not a reversal operation exposed to
      an agent — so it does not upgrade this block above `na`.

cross_links:
  agentic_access: agentic-access/genesco-agentic-access.yml
  conformance: conformance/genesco-conformance.yml
  rate_limits: rate-limits/genesco-rate-limits.yml
  well_known: well-known/genesco-well-known.yml
  llms_txt: llms/genesco-llms.txt