A smart contract is code and data stored at an Ethereum address. A user or another contract can call its functions by submitting a transaction.
Code executes as written
Smart contracts can hold assets and enforce programmed rules, but they cannot correct a bad assumption by themselves. Interactions are generally irreversible, and outside information requires an oracle or another data source.
Public code is not automatically safe code; contract design, testing, upgrade controls, and permissions still matter.
The same call follows the deployed code
Validators execute contract instructions and agree on the resulting state. The contract cannot independently access private systems unless information is supplied through transactions or an oracle.
Deployed does not always mean immutable
Some applications use proxy contracts, administrator keys, or governance to change behavior. Users should check which components can be upgraded and by whom.
Sources
Educational information, not investment advice.