## Problem
The CLI package preserved a generic adapter layer and managed Codex
parser path that did not match the supported customer workflows.
Messaging runs Claude Code sessions, while Codex is supported through
`fcc-codex` and extensions.
## Changes
| Before | After |
| --- | --- |
| `fcc-claude` and `fcc-codex` shared generic adapter plumbing. |
`fcc-claude` and `fcc-codex` use explicit launcher modules. |
| Messaging depended on a generic CLI session abstraction. | Messaging
depends on managed Claude Code sessions. |
| Codex catalog generation lived as a top-level CLI helper. | Codex
catalog generation lives under the Codex launcher owner. |
| Tests asserted deleted internal adapter shapes. | Tests assert
launcher, managed-Claude, and customer-surface behavior. |
## Problem
Codex receives reasoning text from the Responses stream, but FCC does
not emit reasoning-token usage details. Codex reports zero reasoning
tokens even when thinking is present.
## Changes
| Before | After |
| --- | --- |
| Responses usage reported only input, output, and total tokens. |
Responses usage reports input, output, total, and reasoning-token
details when reasoning text exists. |
| Reasoning text had no adapter-owned token estimate. | Reasoning text
gets a Responses-owned best-effort token estimate. |
| Reasoning estimates could exceed reported output tokens. | Reasoning
estimates are capped at reported output tokens. |
| Text-only Responses usage kept the base shape. | Text-only Responses
usage keeps the base shape. |
Codex includes OpenAI tool_search in the tools array; omit it from the Anthropic payload like other passive hosted tools so fcc-codex requests do not fail at conversion.
## Summary
- Split monolithic `core/openai_responses/conversion.py` and `sse.py`
into focused protocol modules (input, output, stream, tools, reasoning,
events, etc.) behind an `OpenAIResponsesAdapter` facade.
- Wire `ClaudeProxyService.create_response()` through the adapter
instead of importing conversion helpers directly, tightening the
API/import boundary.
- Add Codex bridging for Responses `custom_tool_call` items and document
the adapter architecture in `ARCHITECTURE.md`.
## Test plan
- [x] `uv run pytest tests/core/openai_responses/
tests/api/test_openai_responses.py tests/cli/test_adapters.py
tests/contracts/test_import_boundaries.py`
- [x] Full CI via `.\scripts\ci.ps1`
## Summary
- Add `StreamRecoverySession` in
`core/anthropic/stream_recovery_session.py` to centralize early-retry
classification, holdback buffering, retry counting, and flush/discard
behavior shared by Anthropic and OpenAI transports.
- Refactor `AnthropicMessagesTransport` and `OpenAIChatTransport` to use
the shared session instead of duplicating recovery holdback and
early-retry logic.
- Update architecture docs and tests for per-delta overlap trim on
midstream recovery; bump version to 2.2.1.
## Test plan
- [x] `uv run pytest tests/core/anthropic/test_stream_recovery.py`
- [x] `uv run pytest tests/providers/test_anthropic_messages.py
tests/providers/test_streaming_errors.py`
Include copyable upstream status and error bodies in provider SSE failures for OpenAI-compatible and native Anthropic transports, while preserving retry behavior and safe logging defaults.
Update live smoke model defaults for NIM, OpenRouter, and Gemini; normalize tool-call indexes; downgrade DeepSeek forced tool_choice; and add coverage for the provider smoke fixes.
Route these providers through POST /messages with vendor headers and bases (including Kimi model list on OpenAI /v1/models). Remove Z.ai from OpenAI-chat server-tool rejection; extend tests and README.
Alishahryar1