Scored across 20 tools
The buy_* tools have overlapping catalogs (buy_simple and buy_small_pleasure sell the same small_blessing/daily_fortune/pack/window_pick, and hello appears in both buy_signed_record and buy_simple), and three tools (preflight_endpoint, look_at_door, check_before_you_pay) all probe x402 endpoints. The descriptions are unusually explicit about the differences and even say 'either tool is correct,' so an agent can disambiguate with careful reading, but the boundaries are not self-evident from names alone.
Names overwhelmingly follow a snake_case verb-first pattern: buy_*, check_*, read_*, look_*, plus ring_bell, sign_guestbook, verify_artifact, and find_in_catalog. Minor deviations like preflight_endpoint (a check named as a noun) and look_in_window/look_at_door (phrasal verbs) keep it from a perfect 5, but the convention is consistent and predictable.
At 20 tools, the server sits in the heavy 16-25 band. The count is inflated by redundant purchase doors (buy_simple vs buy_small_pleasure) and several overlapping x402 inspection tools that could be consolidated, though no tool is trivially useless.
The storefront covers browse, buy, payment status, order polling, artifact verification, and free store errands, with buy_observation and buy_human_task extending into audits and human fulfillment. Minor gaps remain—no unified purchase-history view and no refund/cancel path—but core workflows have no dead ends.
