SealGuard

Each tool has a describable focus: empty token accounts, token approvals, and a broad risk scan with a safety score. The overlap is real, though — scan_wallet re-reports approvals, permanent delegates, and locked SOL totals that the two list tools already cover in detail, so an agent may hesitate between the aggregate scan and the targeted listings. Descriptions do clarify that scan is the summary view and the list tools are the deep dives.

All snake_case verb_noun names (list_empty_token_accounts, list_token_approvals, scan_wallet) with unambiguous objects. Minor deviation: two tools use the list_ prefix while the third uses scan_, but both are action verbs and the pattern remains readable.

Three tools is lean but justified for a narrowly scoped read-only Solana wallet hygiene checker, with no obvious filler. It sits near the thin end of the range and could arguably fold the two list tools into flagged scan results, but each earns a distinct role.

The surface covers the domain's core inspection needs: empty-account rent recovery, approval/permanent-delegate enumeration, phishing and takeover risk, and reclaimable pump.fun deposits, all summarized by a scored scan. The main gap is that every operation is read-only and remediation is punted to external links, so the server cannot execute the closes or revokes it recommends.