fix: send stream:true on Responses and parse Codex SSE

[felipeinf]

Summary

Codex OAuth chat against /backend-api/codex/responses returns HTTP 400 Stream must be set to true and an SSE body. The Responses path was still sending a unary JSON call (stream omitted) and parsing the whole body as one object.

This sets stream: true on Responses requests, aggregates Codex SSE (response.completed plus output_text.delta / output_text.done when output is empty), and uses the streaming HTTP timeout for that POST.

Related issue

OpenHuman: Closes tinyhumansai/openhuman#5497 (the Stream must be set to true failure). Sibling symptoms in that issue (steering 404, .deb PATH) are out of scope.

Also related: tinyhumansai/openhuman#3201 (same Codex Responses constraint; store=false was already fixed).

API or behavior changes

• Responses-primary OpenAiModel always sends stream: true.
• Unary invoke still returns one ModelResponse; it now accepts either a JSON envelope or aggregated SSE.
• Chat Completions is unchanged.
• Incremental ChatModel::stream on Responses is still a folded unary call (same as before).

Validation

Commands actually run, with their outcome:

• cargo fmt --all -- --check
• cargo clippy --all-targets --all-features -- -D warnings
• cargo build --all-targets --all-features (via clippy + test compile)
• cargo test --all-features (276 lib tests + 1 integration + doctests)

Tests

• SSE response.completed with output_text
• Empty completed.output filled from output_text.done
• Mock TCP server: 400 unless "stream":true, then SSE → ok
• Existing Responses translation test now asserts stream: true

Documentation

No public API docs: this is wire-shape behavior on an existing with_responses_api_primary path.

Checklist

• The change is focused on one logical change
• No new #[allow(...)], #[ignore], or relaxed lints
• No secrets, tokens, or .env contents in the diff or the description

[coderabbitai]

Important

• 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Team

Run ID: 6a040ffa-bba9-4293-938e-d9795aca177e

Warning

Your free Security trial is over. An organization admin can upgrade to Advanced for continuous pull request security review or dismiss this notice.

---

_{Comment @coderabbitai help to get the list of available commands.}

[tinysweeper]

tinysweeper found nothing blocking. Approving.

$0.0000 · 0 in / 0 out

[tinysweeper]

How this change flows

1 changed behaviour across 10 relationships. 6 surrounding behaviours are shown (60 graph nodes walked). 46 further behaviours left out to keep the diagram readable.

flowchart LR
  n0["..._continuation_format_and_provider_options<br/>changed"]:::changed
  n1["Result"]:::impacted
  n2["parse_responses_response"]:::impacted
  n3["translate_responses_request"]:::impacted
  n4["ModelRequest"]:::impacted
  n5["invoke_responses"]:::impacted
  n6["translate_request"]:::impacted
  n0 -->|calls| n3
  n0 -->|tests| n3
  n3 -->|uses| n1
  n3 -->|uses| n4
  n5 -->|uses| n1
  n5 -->|calls| n2
  n5 -->|calls| n3
  n5 -->|uses| n4
  n6 -->|uses| n1
  n6 -->|uses| n4
  classDef changed fill:#0d4429,stroke:#238636,color:#e6edf3
  classDef impacted fill:#161b22,stroke:#6e7681,color:#c9d1d9
  classDef flagged fill:#5a1e02,stroke:#d93f0b,color:#ffffff
  classDef blocking fill:#67060c,stroke:#f85149,color:#ffffff

Loading

_{Green: changed behaviour. Grey: surrounding behaviour. Arrows name the call, use, implementation, or test relationship. Orange: has findings. Red: has a finding that blocks the merge.}