Token Safety (Base)

token_verdict is explicitly documented as running token_risk, rug_check and momentum_score in one call, so three of the five tools overlap heavily and an agent must decide between the aggregate and the individual checks (mainly on price). rug_check vs token_risk are also borderline, separated only by signals versus a composite score. The descriptions do clarify the recommended usage order, keeping it from being worse than a 3.

All names are snake_case with a noun_noun pattern (holder_check, momentum_score, rug_check, token_risk, token_verdict) that reads predictably. There is no verb-based convention, and 'check/score/risk/verdict' are somewhat different framing words, but it is internally readable and not mixed-case.

Five tools is a tight, well-scoped set for a Base token safety server, and each tool maps to a distinct analysis angle. The paused holder_check effectively reduces the live surface to four, which is still reasonable for the purpose.

The surface covers the core lifecycle for evaluating an unknown ERC-20 on Base: risk scoring, rug detection, momentum ranking, and a combined verdict, with holder concentration folded into token_risk. The main gap is that holder_check is paused/dead and there is no token lookup or discovery tool (e.g. resolve symbol to address, list top tokens) beyond momentum ranking.