Prompt caching on Codex was not reliably reusable through the proxy because repeated chat-completions requests could reach the upstream without the same continuity envelope. In practice this showed up most clearly with OpenCode, where cache reads worked in the reference client but not through CLIProxyAPI, although the root cause is broader than OpenCode itself.
The proxy was breaking continuity in several ways: executor-layer Codex request preparation stripped prompt_cache_retention, chat-completions translation did not preserve that field, continuity headers used a different shape than the working client behavior, and OpenAI-style Codex requests could be sent without a stable prompt_cache_key. When that happened, session_id fell back to a fresh random value per request, so upstream Codex treated repeated requests as unrelated turns instead of as part of the same cacheable context.
This change fixes that by preserving caller-provided prompt_cache_retention on Codex execution paths, preserving prompt_cache_retention when translating OpenAI chat-completions requests to Codex, aligning Codex continuity headers to session_id, and introducing an explicit Codex continuity policy that derives a stable continuity key from the best available signal. The resolution order prefers an explicit prompt_cache_key, then execution session metadata, then an explicit idempotency key, then stable request-affinity metadata, then a stable client-principal hash, and finally a stable auth-ID hash when no better continuity signal exists.
The same continuity key is applied to both prompt_cache_key in the request body and session_id in the request headers so repeated requests reuse the same upstream cache/session identity. The auth manager also keeps auth selection sticky for repeated request sequences, preventing otherwise-equivalent Codex requests from drifting across different upstream auth contexts and accidentally breaking cache reuse.
To keep the implementation maintainable, the continuity resolution and diagnostics are centralized in a dedicated Codex continuity helper instead of being scattered across executor flow code. Regression coverage now verifies retention preservation, continuity-key precedence, stable auth-ID fallback, websocket parity, translator preservation, and auth-affinity behavior. Manual validation confirmed prompt cache reads now occur through CLIProxyAPI when using Codex via OpenCode, and the fix should also benefit other clients that rely on stable repeated Codex request continuity.
When assistant has tool_calls but no text content, the translator
emitted an empty message into the Responses API input array before
function_call items. The API then couldn't match function_call_output
to its function_call by call_id, returning:
No tool output found for function call ...
Only emit assistant messages that have content parts. Tool-call-only
messages now produce function_call items directly.
Added 9 tests for tool calling translation covering single/parallel
calls, multi-turn conversations, name shortening, empty content
edge cases, and call_id integrity.
The instructions restore logic was originally needed when the proxy
injected custom instructions (per-model system prompts) into requests.
Since ac802a46 removed the injection system, the proxy no longer
modifies instructions before forwarding. The upstream response's
instructions field now matches the client's original value, making
the restore a no-op.
Also removes unused sjson import.
Closesrouter-for-me/CLIProxyAPI#1868
The OpenAI Chat Completions translator was silently dropping
response.function_call_arguments.delta and
response.function_call_arguments.done Codex SSE events, meaning
tool call arguments were never streamed incrementally to clients.
Add proper handling mirroring the proven Claude translator pattern:
- response.output_item.added: announce tool call (id, name, empty args)
- response.function_call_arguments.delta: stream argument chunks
- response.function_call_arguments.done: emit full args if no deltas
- response.output_item.done: defensive fallback for backward compat
State tracking via HasReceivedArgumentsDelta and HasToolCallAnnounced
ensures no duplicate argument emission and correct behavior for models
like codex-spark that skip delta events entirely.
- Replaced all instances of `bytes.Clone` with direct references to enhance efficiency.
- Simplified payload handling across executors and translators by eliminating unnecessary data duplication.
- Added conditional logic for Codex instruction injection based on configuration.
- Updated role terminology from "user" to "developer" for better alignment with context.
- Added logic to transform `inputResults` into structured JSON for improved processing.
- Removed redundant `safety_identifier` field in executor payload to streamline requests.
- Pass through non-function tool definitions like web_search
- Translate tool_choice for built-in tools and function tools
- Add regression tests for built-in tool passthrough
Replace the "~<n>" suffix with "_<n>" when generating unique short names in codex translators (Claude, Gemini, OpenAI chat).
This avoids using a special character in identifiers, improving compatibility with downstream APIs while preserving length constraints.
fix(translator): consolidate temperature and top_p conditionals in OpenAI Claude request
Fixed: #169
fix(translator): adjust instruction strings in Codex Claude and OpenAI responses
docs: add GPT-5 Codex guidelines for CLI usage
- Added detailed guidelines for GPT-5 Codex in Codex CLI.
- Expanded instructions on sandboxing, approvals, editing constraints, and style requirements.
- Included presentation and response formatting best practices.
fix(codex_instructions): update comparison logic to use prefix matching
- Changed system instructions comparison to use `strings.HasPrefix` for improved flexibility.
- Added comprehensive instructions for Codex CLI harness, sandboxing, approvals, and editing constraints to `internal/misc/codex_instructions/`.
- Clarified `approval_policy` configurations and scenarios requiring escalated permissions.
- Provided detailed style and structure guidelines for presenting results in the Codex CLI.
refactor(translator): streamline Codex response handling and remove redundant code
- Updated `ConvertCodexResponseToOpenAIResponses` logic for clarity and consistency.
- Simplified `ConvertCodexResponseToOpenAIResponsesNonStream` by removing unnecessary buffer setup and scanner logic.
- Switched to using `sjson.SetRaw` for improved processing of raw input strings.
feat(translator): add support for single input string in Codex responses parser
- Modified input parsing logic to handle cases where input is a single string instead of an array.
- Added functionality to convert single string inputs into structured JSON format.
- Added new "Gemini 2.5 Flash Image Preview" model definition, with enhanced image generation capabilities.
- Increased scanner buffer size to 20,971,520 bytes across executors and translators to handle larger payloads.
The OpenAI Codex Responses API (chatgpt.com/backend-api/codex/responses)
rejects requests containing max_output_tokens and max_completion_tokens fields,
causing Factory CLI to fail with "Unsupported parameter" errors.
This fix strips these incompatible fields during request translation, allowing
Factory CLI to work properly with CLIProxyAPI when using ChatGPT Plus/Pro OAuth.
Fixes compatibility issue where Factory sends token limit parameters that aren't
supported by the Codex Responses endpoint.
