Skip to main content

Recover recorded State alongside your workflow

Kam State adds versioned application state to Amazon Bedrock AgentCore.

Use Kam State’s retained revisions and recorded reconciliation progress when an AgentCore workflow resumes, keeping runtime continuation with your application.

Internal implementationBackend review 2026-10-14 · prelaunchRead this guide as Markdown
Guide context

Review Kam State recovery alongside Amazon Bedrock AgentCore execution. Retained revisions and durable reconciliation progress reduce custom recovery bookkeeping for accepted State; workflow continuation remains separate.

When an AgentCore application resumes work, read the Kam State evidence it needs again and recheck applicability. Kam retains accepted revisions and its own reconciliation progress, reducing the recovery bookkeeping teams must build for accepted State. Runtime continuation, tool retries and the decision to act remain with the runtime and application. Managed operational checkpoints remain planned.

On this page · 3 sections

Rejoin execution with recorded State

  1. Resume the workload

    Your AgentCore runtime and application own continuation and tool attempts.

  2. Read and revalidate State

    Kam retains admitted versions and recovers its own reconciliation progress.

  3. Stop on uncertainty

    Your application must verify unknown external effects and current permissions before deciding whether to retry.

Two timelines should remain independent#

Your AgentCore application owns how execution resumes. Kam preserves accepted State and advances supported reconciliation as recorded evidence changes. On the next turn, the application must read again and check whether a previous Observation still applies.

Keep workflow continuation in the runtime and application. Kam supplies exact State references and its own recorded processing progress; neither proves that an external tool action completed.

Use exact references for checkpoints#

The planned operational checkpoint is a durable coordinate across exact Snapshot or generation, View, Active State, Observation, source fences and computational versions. Immutable referenced objects can be structurally shared across many checkpoints.

Exact references make retained versions inspectable without copying every payload. Replay does not rerun tools, prove outside actions or remove retention limits.

Durable reconciliation preserves processing progress#

Affected-View discovery persists bounded progress and can redrive lost wakeups. Conditional commits recheck lifecycle, membership and source-State versions so delayed work cannot overwrite newer authority.

The future Turn layer should carry the same discipline into model submission and effects: explicit UNKNOWN outcomes, idempotent stage receipts, and revalidation instead of blind retries.

Agent-readable documentation manifest