Skip to content

feat(runtime/codex): multimodal I/O support — accept inbound images (#398)#528

Merged
alexey-pelykh merged 1 commit intomainfrom
feat/codex-multimodal-image-input
Mar 9, 2026
Merged

feat(runtime/codex): multimodal I/O support — accept inbound images (#398)#528
alexey-pelykh merged 1 commit intomainfrom
feat/codex-multimodal-image-input

Conversation

@alexey-pelykh
Copy link
Copy Markdown

Summary

  • Update mediaCapabilities to declare acceptsInbound: ["image/"] (upstream --image + --json bug is fixed)
  • Append --image <filePath> flags in buildArgs() for each image MediaAttachment
  • Skip images on session resume (exec resume) since Codex propagates conversation context internally
  • Add 9 unit tests (7 buildArgs image scenarios + 2 mediaCapabilities assertions)
  • Add live smoke test for image description

Closes #398

Test plan

  • Unit tests pass (60 tests, including 9 new)
  • Type-check clean (no new errors)
  • Formatter/linter clean
  • CI build + test pass
  • LIVE=1 pnpm test:live smoke test (requires Codex CLI)

🤖 Generated with Claude Code

…398)

Enable the Codex CLI runtime to accept image attachments by updating
mediaCapabilities to declare image/ support and appending --image flags
in buildArgs() for each image MediaAttachment with a filePath. Images
are skipped on session resume since Codex propagates context internally.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
@alexey-pelykh alexey-pelykh enabled auto-merge (squash) March 9, 2026 09:59
@alexey-pelykh alexey-pelykh merged commit 3d3c11a into main Mar 9, 2026
7 checks passed
@alexey-pelykh alexey-pelykh deleted the feat/codex-multimodal-image-input branch March 9, 2026 10:07
@alexey-pelykh alexey-pelykh restored the feat/codex-multimodal-image-input branch March 9, 2026 10:12
@alexey-pelykh alexey-pelykh deleted the feat/codex-multimodal-image-input branch March 9, 2026 10:21
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

feat(runtime/codex): multimodal I/O support

1 participant