The CLI does one thing on the page

usage: evergreen scan <contract-id> [<contract-id> ...] [--keys-file <path> | --no-data-keys] [--require-declared-scope] [--threshold N] [--json] [--cost [--ledgers N]] [--optimize] Reads instance/Wasm and supplied persistent/temporary keys on Stellar Testnet. Keys file: { "dataKeys": ["base64 XDR LedgerKey", ...] } PASS SEVERAL CONTRACTS TOGETHER to see real shared-code blast radius. Contracts built from the same Wasm share ONE ContractCode ledger entry, and a scan of one contract cannot tell whether others depend on it — the chain does not index reverse dependencies from a single query, so that entry reports "sharing undetermined". Naming them together resolves it: evergreen scan <A> code entry: 1 consumer, sharing UNDETERMINED evergreen scan <A> <B> <C> code entry: 3 consumers, SHARED, they fail together Every scan prints which contracts it actually scanned, so a mistyped or dropped argument is visible rather than inferred. --keys-file takes exactly one contract, because data keys belong to a specific contract and the file does not say which. Exit: 0 everything scanned is healthy; 1 observed low TTL; 2 error; 3 the scan came back incomplete (entry missing, TTL unavailable, executable not followable, or nothing observed). Precedence: 2 > 3 > 1 > 0. Exit status never authorizes a transaction. Scanning reads the keys it is given; it cannot enumerate a contract's storage, so a clean exit means "everything I was asked to check is healthy" and never "this contract is fully healthy". Coverage is printed with every scan. --threshold N act-now threshold in LEDGERS, default 17,280 (~1 day). What evergreen-check sets in CI: a repository that wants a week of warning fails its build at 120,960, not at ours. Both health tiers move with it — WARNING widens as the action threshold rises, so raising it never silently narrows the earlier warning. Exit 1 means an entry is at or below this value; the boundary is inclusive, because remaining exactly N is already the margin you set out to keep. Changes what is REPORTED and never what is written. --no-data-keys assert this contract has no data keys beyond its instance. Only its author can know that; it is a caller declaration and is never independently verified. --require-declared-scope also exit 3 when scope was not declared. Intended for CI on a contract you own; evergreen-check sets it by default. --json machine-readable output. The human view is a summary; JSON is the complete record, including every issue. --optimize append conditional storage advice with evidence and scope limits. Reads network minimum lifetimes; no payer needed. Add --cost for current rent quotes. No storage is changed. --cost [--ledgers N] estimate what extending every entry by N more ledgers would cost, priced by simulating against the network. Default N is 518,400 (~30 days). Nothing is submitted. "--ledgers N" means "give me N MORE ledgers". The protocol wants an absolute target, so the CLI computes it for you and caps it at max_entry_ttl - 1, saying so when it does. Costs are estimates: rent pricing varies with network state and has differed ~18% between days. Health states, printed per entry and as a worst-of summary: HEALTHY above threshold. WARNING low, recoverable, and affects only this contract. CRITICAL expired, OR temporary (deleted at expiry, unrecoverable), OR low and SHARED — a code entry shared by N contracts at 3 days is N contracts at 3 days, not one. UNKNOWN TTL could not be read. Not healthy; unread. Colour is added only for an interactive terminal and honours NO_COLOR. The state word always prints, so piped output and screenshots lose nothing.