Skip to content

[P2-02] Add voice, realtime, and audio streaming support #184

Description

@erikhoward

Goal

Extend PetalFlow beyond text/image/file workflows for realtime and audio agent experiences.

Scope

  • Define audio input/output artifacts and realtime event semantics.
  • Support session lifecycle, interruption, backpressure, cancellation, and reconnect.
  • Add provider capability negotiation and latency/quality telemetry.
  • Define privacy, retention, and content-type limits for audio data.

Acceptance criteria

  • A realtime workflow can start, stream, interrupt, and close a session deterministically.
  • Audio payloads follow the artifact lifecycle and authorization policy.
  • Provider-specific features do not leak into the core contract without capability checks.
  • Reconnect, cancellation, malformed media, and backpressure are tested.

Priority

P2 — valuable post-1.0 capability.

See the PetalFlow 1.0.0 Readiness Report for the audit context.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions