Skip to content

feat(ai): define a local inference adapter and deterministic test backend #81

Closed

Description

@McanKul

Separate OffPDF product code from the selected inference implementation.

Scope

  • Define a small interface for load, unload, generate, cancel, status, and health checks.
  • Map runtime failures into normal structured app errors.
  • Provide a deterministic fake backend for frontend and integration tests.
  • Keep model acquisition and document processing outside this layer.

Acceptance criteria

  • Product features do not import a concrete runtime directly.
  • Cancellation and failure paths are testable without a real model.
  • The adapter exposes no network or arbitrary-command capability.

Activity

  1. added
    enhancementNew feature or request
    help wantedExtra attention is needed
    area: aiOptional on-device AI and document intelligence
    on Aug 30, 2026
  2. McanKul commented on Sep 10, 2026

    @McanKul
    OwnerAuthor

    Completed in #103 and merged into epic/local-ai-runtime.

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

    area: aiOptional on-device AI and document intelligenceenhancementNew feature or requesthelp wantedExtra attention is neededpriority: highHigh-priority roadmap worktarget: local-ai-runtimeOpen the pull request against epic/local-ai-runtime

    Projects

    No projects

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions