From one transaction to living in your workflow.
Two things ship today: local transaction analysis, and confidentiality analysis for Move. From there the same engine grows into your workflow — your editor, your tests, your CI, your network, and your stakeholders — adding private static insight and results from real encrypted simulations. And as software begins to sign and pay on its own, it becomes the check that runs first: reading what a transaction really does, not what a prompt claimed, before it is signed.
Multi-chain transaction analysis
Analyze in-browser or in your workflow any transaction on Sui, Aptos, or Movement (mainnet or testnet) by digest, hash, version, or JSON. For Sui: Kinetics builds the dataflow graph, computes dependency depth, tracks taint from inputs to sinks, checks linear-resource conservation, and attributes gas. For Aptos and Movement: it decodes the payload, balance movements, events, and write-set. Public simulation is live alongside it: dry-run a transfer against current chain state and read the balance changes and gas it would produce, with nothing signed and nothing broadcast. It is the same read an agent's signer, or a wallet, should run before approving anything: the payload's own content, not the label the calling app gave it.
Confidentiality analysis for Move
Kinetics now reads a Move contract's confidentiality surface the same way it reads a transaction statically in your browser or local workflow before you deploy. For Sui it reads a package's Seal access policy: every seal_approve function is placed as Open, Restricted, or Locked, and an Open policy (*e.g.*, one that hands a decryption key to any caller) is flagged, alongside the convention footguns (a policy that is not an entry function, that omits the id: vector<u8> identity, or that is not side-effect free). For Aptos it reads a module's use of Confidential Assets: it separates moving an encrypted amount from disclosing it, and traces the paths that put a confidential amount back in the open (*e.g.*, an amount emitted in an event, a decryption key handed to a sink, the public amounts at a deposit or a withdraw).
Every run reports the exposure surface, the findings with a concrete fix for each, and a three-observer view of what the public ledger, an authorized viewer, and an unintended observer each end up holding. It runs entirely client-side; no source ever leaves the browser. It parses Move source code into an AST: reachability follows calls into helpers and information flow is traced through the types. A lexical scan is also maintained as a fallback. The underlying engine ships in the @choosek/kinetics library, so the same rules are applied in the browser, on the command line, or as part of a CI workflow on GitHub Actions.
Complete SDK, CI integration, and private simulations
Integrate analysis into your tests and CI. Assert no dangling results, a bounded dependency depth, or that only expected inputs reach transfer sinks. Gate a pull request on any confidentiality leak, exactly as a leak in a seal_approve policy would. Then private simulation: the counterfactual dry-runs you can already run in the browser, brought into private environments where your inputs and outputs stay encrypted, enforced by Move policies on Sui's Seal framework. Those same assertions double as an agent's policy (only specific recipients, spend within a cap) so an x402-style payment can be checked against it before signing.
Virtual testnets and cross-chain support
Fork mainnet state into isolated environments in seconds. Leverage full support for virtual testnets across Sui, Aptos, and Movement. Deploy and test against real liquidity before touching mainnet. The same forks give an agent a place to rehearse a proposed action against real state before it ever signs one.
Safety proofs and enterprise features
Generate and share cryptographic proofs that a contract passes defined safety and confidentiality properties, so audit confidence can be shared without revealing implementation details. The same machinery can prove an agent's policy holds over every transaction it could produce: it is not over its cap, outside the permissions granted to it, or being steered into a drain. In addition, enjoy all the controls teams need to run Kinetics across an organization.