Standard
  • Docs
  • Examples
  • Pricing
  • About

Get started

  • Overview
  • How Standard works
  • Run your first task
  • Connect
  • Join a Space

Build

  • Examples
  • Operation profiles
  • Integration reference

Concepts

  • Inputs and permissions
  • Accounts and billing
  • Results and lifecycle
  • Verification checks
  • Follow-on tasks
  • Cleanup

Security

  • Security model
Documentation menu

Get started

  • Overview
  • How Standard works
  • Run your first task
  • Connect
  • Join a Space

Build

  • Examples
  • Operation profiles
  • Integration reference

Concepts

  • Inputs and permissions
  • Accounts and billing
  • Results and lifecycle
  • Verification checks
  • Follow-on tasks
  • Cleanup

Security

  • Security model
Standard

Managed confidential execution for sensitive functions.

© 2026 Standard Runtime, Inc. All rights reserved.

  • Security
  • Docs
  • Examples
  • Pricing
  • FAQ
  • About
  • Contact
  • Terms
  • Privacy
  • Cookies
  • Docs
  • Examples
  • Pricing
  • About
  • Get started

    • Overview
    • How Standard works
    • Run your first task
    • Connect
    • Join a Space

    Build

    • Examples
    • Operation profiles
    • Integration reference

    Concepts

    • Inputs and permissions
    • Accounts and billing
    • Results and lifecycle
    • Verification checks
    • Follow-on tasks
    • Cleanup

    Security

    • Security model

    Get started

    • Overview
    • How Standard works
    • Run your first task
    • Connect
    • Join a Space

    Build

    • Examples
    • Operation profiles
    • Integration reference

    Concepts

    • Inputs and permissions
    • Accounts and billing
    • Results and lifecycle
    • Verification checks
    • Follow-on tasks
    • Cleanup

    Security

    • Security model

    Results and lifecycle

    Automatic verification and result delivery, approval requests, failures, retries and cleanup status.

    A result becomes useful to the next agent when its execution evidence is accepted and its contents satisfy the next step's policy. Standard reports progress and cleanup separately so the workflow can distinguish completion, rejection, and unresolved state.

    Automatic verification

    The receiving client checks the evidence before it delivers a result; with the SDK, that check runs in your environment. Routine work does not need anyone to inspect signatures, bindings or evidence by hand. Verification checks is the evidence reference: each check the client runs, for integrators and auditors who want to inspect them independently.

    From request to result

    The integration reports submission, any outstanding authority requirements, execution, verification, and the resulting completion or failure state. Follow the statuses and error classes of the selected operation; a submitted request or running process is not evidence of a completed result.

    Protected execution applies the release policy and encrypts the permitted result for its recipient. The receiving integration checks the associated evidence against accepted references before returning a verified result.

    Verify evidence, then evaluate the outcome

    EvidenceReported resultNext step
    AcceptedRequired application checks passContinue if the next action is authorized.
    AcceptedA test or security check failsBlock release and use the findings for remediation.
    Missing or invalidAny reported outcomeReject the handoff; do not rely on the reported success.

    In the release example, the release agent checks the artifact digest as well as the test and scanner results. Evidence for one candidate cannot justify releasing a different artifact.

    Outstanding authority

    Routine requests use standing permissions. When a request needs authority that has not been supplied, execution or release stays blocked. The policy may allow an authorized decision to resolve that state; it cannot waive a failed evidence check by labeling the result successful.

    Failures and uncertain outcomes

    Missing permission, refused key access, input mismatches, unsupported formats, and failed verification must remain visible to the caller. A protected operation must not fall back to unprotected processing to complete the request.

    Use the operation's status and correlation information to distinguish a rejected request from a result that is still being resolved. A lost response does not establish whether an external effect completed.

    Retries

    PreviousAccounts and billingNextVerification checks

    On this page

    • Automatic verification
    • From request to result
    • Verify evidence, then evaluate the outcome
    • Outstanding authority
    • Failures and uncertain outcomes
    • Retries
    • Cleanup and retention

    Use the integration's supported idempotency and status mechanisms. Retry only failures identified as transient. Correct the cause of a permission, input, compatibility, or verification failure before submitting again.

    For an operation involving an external effect, reconcile its status before repeating it. An execution record does not create an exactly-once guarantee for a third-party service.

    Cleanup and retention

    Cleanup has a separate status. Keep a pending or unconfirmed cleanup state distinct from a confirmed one. Original stored inputs, released results, and retained evidence follow their respective retention rules.

    • Verification checks
    • Follow-on tasks
    • Cleanup