Deliver generated images through channel and web abstractions #24

Closed
opened 2026-10-27 13:34:38 +00:00 by sam · 0 comments
Owner

Scope

Extend the existing message/result delivery abstraction so generated images can reach users safely through supported channels and the web UI. Preserve useful metadata such as prompt/provider/model and errors without exposing credentials or internal filesystem paths.

Use bounded temporary storage/cleanup and a clear policy for formats, dimensions, compression, and maximum response size. Ensure a failed upload does not leave the user with a success-looking text response.

Acceptance criteria

  • A provider result can be delivered as an image through each currently supported delivery path, or unsupported paths report a clear limitation.
  • Temporary files and downloaded inputs have bounded size and deterministic cleanup.
  • MIME/format handling, alt text/caption, and metadata are tested.
  • Delivery failures, retries/cancellation, and partial-success behavior are tested.
  • Existing text and file delivery behavior remains unchanged.

Design

See docs/prds/image-capability-contract.md#delivery-through-channels-and-web-501 for the settled design: this reuses generate_video's existing MultimodalResult -> MediaEvent pipeline (Telegram already works), so the only new work is the generate_image/edit_image tool wrapper and its bootstrap.go registration.

Suggested /council lenses

lens-deletionist, lens-contract -- run /council --lenses deletionist,contract once gate-clean, before opening the PR.

Close reason

Closed


Imported from beads archie-core-1786748942243-6-f109697e · GitHub #501 · type task · priority P2 · created 2026-08-14 · closed 2026-10-08

## Scope Extend the existing message/result delivery abstraction so generated images can reach users safely through supported channels and the web UI. Preserve useful metadata such as prompt/provider/model and errors without exposing credentials or internal filesystem paths. Use bounded temporary storage/cleanup and a clear policy for formats, dimensions, compression, and maximum response size. Ensure a failed upload does not leave the user with a success-looking text response. ## Acceptance criteria - A provider result can be delivered as an image through each currently supported delivery path, or unsupported paths report a clear limitation. - Temporary files and downloaded inputs have bounded size and deterministic cleanup. - MIME/format handling, alt text/caption, and metadata are tested. - Delivery failures, retries/cancellation, and partial-success behavior are tested. - Existing text and file delivery behavior remains unchanged. ## Design See docs/prds/image-capability-contract.md#delivery-through-channels-and-web-501 for the settled design: this reuses generate_video's existing MultimodalResult -> MediaEvent pipeline (Telegram already works), so the only new work is the generate_image/edit_image tool wrapper and its bootstrap.go registration. ## Suggested /council lenses lens-deletionist, lens-contract -- run `/council --lenses deletionist,contract` once gate-clean, before opening the PR. ### Close reason Closed --- _Imported from beads `archie-core-1786748942243-6-f109697e` · GitHub #501 · type task · priority P2 · created 2026-08-14 · closed 2026-10-08_
sam added the mediumtask labels 2026-10-27 13:34:38 +00:00
sam self-assigned this 2026-10-27 13:34:38 +00:00
sam closed this issue 2026-10-27 13:34:38 +00:00
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: sam/archie#24