sl-entity-infrastructurelisted

# Entity infrastructure ## TL;DR | Layer | Question it answers | Effort | Impact | |---|---|---|---| | **Crawler access** | Is the assistant allowed in at all? | Minutes | Blocking. Nothing else matters if this fails | | **Bing index** | Can ChatGPT search see the site? | Hours | Blocking for one whole assistant | | **Schema depth** | Does the machine know what this is? | Days | High | | **Entity graph** | Can it be disambiguated from similar names? | Weeks | High, and durable | | **`llms.txt`** | Emerging convention, unproven | Minutes | Low. Do it last, claim nothing | **Order matters.** Work top down. The first two are cheap and blocking; the fourth is slow and compounding. > **`freshness: volatile`.** Crawler user-agent names, `llms.txt` adoption and knowledge-panel > behaviour all change without notice. Re-verify every user-agent string and every claim in this > file against current platform documentation before acting on it. ## Before starting 1. **Load [`CONTEXT.md`](../../CONTEXT.md).** 2. **Confirm the symptom is identity, not content.** Ask the assistants directly: *"What is [brand]?"* Three distinct failures, three different remedies: | Response | Failure | Fix | |---|---|---| | "I do not have information about that" | Not resolvable | This skill, layers 1 to 4 | | Describes a **different** company with a similar name | Disambiguation | This skill, layer 4 first | | Describes it correctly but never cites it | Not an identity problem | `sl-citation-content