audit_page performs the scoring action on a URL, while explain_check only describes a single element's criteria; there is no plausible way to confuse the two. Each tool also has a distinct return shape (score plus evidence vs. explanatory text).

Both names use the same snake_case verb_noun pattern (audit_page, explain_check), with clear, concrete verbs. No mixed conventions or vague catch-all names.

Two tools is on the thin side, though they map cleanly to the only two actions implied by the domain (score a page, explain a criterion). The surface feels borderline minimal rather than over- or under-built for a single-purpose auditor.

The core lifecycle—run an audit and look up what each of the 20 checks means—is covered, and audit_page returns evidence plus fixes for every element, so there are few dead ends. A minor gap: there is no tool to enumerate e01...e20, so an agent must obtain IDs from audit output or guess them.