* feat(codex): support native Anthropic Messages protocol as upstream
Allow gateways that only expose the native Anthropic Messages protocol
(/v1/messages) to be used by Codex: the local proxy performs bidirectional
request/response/streaming conversion between Responses and Anthropic.
Backend:
- Add two conversion modules: transform_codex_anthropic / streaming_codex_anthropic
- codex.rs: add routing detection and auth: ANTHROPIC_AUTH_TOKEN→Bearer (default),
ANTHROPIC_API_KEY→x-api-key, mutually exclusive
- handlers.rs: add handle_codex_anthropic_to_responses_transform
- forwarder.rs: support optional Claude Code client fingerprint impersonation
(User-Agent / anthropic-beta / x-app / system prompt first-line injection)
and /responses→/v1/messages rewriting
- codex_config: the anthropic format reuses the NativeResponses profile to strip
custom tools
- ProviderMeta: add impersonateClaudeCode
Frontend:
- CodexApiFormat: add "anthropic"; the form adds auth field selection and an
impersonation toggle
- Add en/ja/zh/zh-TW copy
Robustness:
- Downgrade when tool history / forced tool_choice conflicts with extended
thinking, avoiding upstream 400s
- Emit cache_creation_input_tokens in usage and use saturating_add to guard
against overflow
- Append a unique suffix to non-streaming/streaming output-item ids to avoid
multi-segment text/thinking overwriting each other
* fix(codex): harden Anthropic bridge against empty text blocks and truncated streams
- Drop empty/whitespace-only text content blocks when rebuilding Anthropic
messages from Responses history; Anthropic 400s on empty text blocks (e.g. an
empty assistant text emitted alongside a tool_use), which broke follow-up and
tool-result requests. Also drop messages left without content.
- Do not report a truncated Anthropic stream as completed: when the SSE
connection ends before message_stop with no stop_reason, emit an incomplete
response if partial output exists, or a failed (stream_truncated) response
otherwise, mirroring the chat converter's EOF handling.
* fix(codex): resolve Anthropic-bridge review findings
Blocking:
1. Defer stripping the [1m] long-context marker until after catalog matching and model write-back, re-stripping it on the final Codex→Anthropic body and setting a flag to emit the context-1m-2025-08-07 beta header, so the marker is no longer lost or overridden by the default model.
2. Gate Anthropic thinking on the trailing turn only (via trailing_turn_allows_thinking) instead of scanning full history, so a Codex session that resends history each round no longer permanently loses thinking after the first tool call.
3. Inject 5m ephemeral cache_control on the Codex→Anthropic body by reusing cache_injector (handling the system string→array conversion), so system/tools/history are cached instead of re-sent at full price every round.
4. Add a shared base_url_is_full_endpoint helper (normalizing whitespace/query/fragment/trailing slash) used by both the Anthropic and Chat paths, so a base URL already ending in /v1/messages is treated as a full endpoint instead of double-appending to /v1/messages/v1/messages.
5. Align the catalog tool-profile predicate with the routing predicate so resolve_codex_catalog_tool_profile returns the Anthropic profile whenever the request converts to Anthropic, preventing freeform tools like apply_patch from being silently filtered by a ProxyChat catalog.
6. Explicitly disable native web_search for the Anthropic profile (including the no-catalog branch) via set_codex_native_web_search_field, so Codex no longer treats it as available while the transform silently drops it.
7. Only forward tool_choice when tools survive filtering, dropping it otherwise, to avoid a non-retryable 400 from upstream when tool_choice is sent with no tools.
8. Lower the fallback default to max_tokens=8192 (only when max_output_tokens is omitted) and clamp the thinking budget to max_tokens/2, disabling thinking below the 1024 floor, to avoid hard 400s on low-output-ceiling models/gateways.
Minor:
9. Centralize the Codex/OpenAI fingerprint-header denylist in is_codex_client_fingerprint_header so impersonating Claude Code uniformly drops originator/session_id/conversation_id/chatgpt-account-id/x-client-request-id/openai-* and the x-stainless-*/x-codex-* prefixes.
10. Retain content_block_start.input as start_input and fall back to it at block close when no input_json_delta arrived, so a gateway that carries the full tool input on the start event no longer yields empty tool arguments.
11. Extract a shared codex_responses_sse module as the single Responses SSE envelope builder that both the chat and anthropic streaming emitters delegate to, with byte-for-byte-unchanged wire output, so future event-format fixes touch one place instead of two.
* fix(codex): add per-provider max_output_tokens override for Codex→Anthropic path
Codex does not forward model_max_output_tokens in the request body,
causing the proxy to fall back to a conservative 8192 default. This
truncates long or thinking-heavy responses (stop_reason=max_tokens).
- Add maxOutputTokens field to ProviderMeta (Rust + TypeScript)
- Inject the value into the Anthropic request body before transform,
taking precedence over request-supplied and default values
- Add numeric input in Codex form fields (Anthropic format only)
- Add i18n entries for label, placeholder, and hint (en/zh/zh-TW/ja)
- Include roundtrip and omission unit tests for the new field
* fix(codex): harden Anthropic protocol bridge
* fix(codex): address Anthropic bridge review
* fix(codex): preserve flattened Anthropic inputs
* fix(codex): harden Anthropic recovery paths
* fix(ci): satisfy Rust clippy
---------
Co-authored-by: Jason <farion1231@gmail.com>
CI cargo fmt --check failed on transform_codex_chat.rs (the wrapping
introduced upstream did not match rustfmt 1.95.0). Apply cargo fmt so the
chat_legacy_function_call_to_response_item call uses the expected layout.
* Chat API: skip tool calls with missing function names
Some providers send empty or absent function names in streaming
tool call deltas. Previously these produced invalid output items.
- Don't overwrite accumulated state.name with empty deltas
- Skip tool calls that never received a valid name (instead of
falling back to 'unknown_tool')
- Apply the same defensive guard in finalize_tools and the
non-streaming path
* Address review: defer empty-name skip to finalization
Require both call_id and name before triggering should_add,
instead of skipping eagerly when name is absent in the first
delta. This handles providers that send id before name,
as suggested in the Codex review.
* Guard legacy function_call against empty name
Return Option<Value> from chat_legacy_function_call_to_response_item,
returning None when function_call.name is missing or empty. This covers
the legacy message.function_call path that the original guard missed.
* Remove unreachable unknown_tool fallback
---------
Co-authored-by: Jason <farion1231@gmail.com>
* fix(codex): restore cached tool call fields
* refactor(codex): merge duplicate enrich loops in chat history
enrich_call_item_from_cache copied the fill-if-empty loop for
reasoning_content/reasoning. The two loops are identical and key
order is irrelevant, so fold both key sets into a single loop.
Pure refactor, no behavior change; codex_chat_history tests pass.
---------
Co-authored-by: Jason <farion1231@gmail.com>
DeepSeek's Anthropic-compatible endpoint rejects requests where
thinking.type=disabled coexists with effort parameters, returning
HTTP 400. This breaks Claude Code 2.1.166+ sub-agents (Workflow/Dynamic
Workflow), which hardcode thinking:disabled.
Rather than overriding thinking:disabled, remove the conflicting effort
parameters (output_config.effort / reasoning_effort) to respect Claude
Code's intent — sub-agents don't need to display reasoning.
Fixes: https://github.com/deepseek-ai/DeepSeek-V3/issues/1397
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The provider panel health check sent a real streaming model request, which many third-party providers block (401/403/WAF), causing false negatives while only stable official endpoints passed. Replace it with a lightweight reachability probe: GET the provider base_url and treat any HTTP response (200/4xx/5xx) as reachable; only DNS/connect/TLS/timeout count as failure. Latency is the probe's TTFB.
Backend (services/stream_check.rs): rewrite ~2200 -> ~350 lines, dropping real-request building, format conversion, auth and API-path resolution while keeping per-app base_url extraction. Defaults: 8s timeout, 1 retry, 1500ms degraded threshold.
Failover invariant: the reachability check must never reset the circuit breaker (reachable != usable; a 403 host is reachable but broken for real traffic). Remove the resetCircuitBreaker call from useStreamCheck; failover failure detection stays driven solely by real proxy traffic (forwarder/circuit_breaker untouched). useResetCircuitBreaker is kept dormant for a future manual-recovery entry.
Open the check to all providers: drop the official/copilot/codex-oauth/third-party gating and the 'sends a real request' confirm dialog. For official providers whose base_url is intentionally empty, fall back to the endpoint the client actually uses (Claude -> api.anthropic.com, Codex -> chatgpt.com/backend-api/codex, Gemini -> generativelanguage). Non-official providers with a missing base_url still error to avoid a false green light. Claude Desktop Official is native 1P mode (talks to claude.ai, cc-switch not in the request path, no reliable endpoint) so its button stays hidden.
Slim StreamCheckConfig and per-provider testConfig to timeout/threshold/retries (drop test model + prompt); sync zh/en/ja/zh-TW. Retain the now-unused anthropic_to_openai/anthropic_to_gemini transform utilities and their test suites.
Builds on #2774 (which fixed cache_read for the streaming openai_chat path).
Two gaps remained, both double-counting cache tokens when a Claude client
meters as app_type="claude" (input_includes_cache_read=false):
1. cache_read was still added to input on the non-streaming openai_chat path
(transform.rs openai_to_anthropic) and the whole openai_responses family
(transform_responses.rs build_anthropic_usage_from_responses, covering the
non-streaming call site and both streaming_responses call sites).
2. cache_creation was never subtracted on any converted path, including the
streaming openai_chat path #2774 had already touched. Claude billing treats
cache_creation as a separate bucket, so an inclusive upstream carrying a
direct cache_creation_input_tokens field billed it twice.
All four metering points now compute:
input = prompt_tokens - cache_read - cache_creation
restoring the invariant input + cache_read + cache_creation == prompt_tokens.
Pure OpenAI upstreams are unaffected (no cache_creation concept/field).
Tests: update direct-cache assertions (40->20), add a streaming conservation
regression test, and pin prompt<cache underflow (saturating clamp to 0) for all
three metering functions. cargo test 1573 pass, clippy clean.
Note: fix is forward-only; historical rows are not recomputed (cost is frozen at
log time and app_type="claude" mixes native + converted rows).
Audited all proxy format-conversion paths (Chat<->Message, Chat<->Response,
Gemini<->Message) for usage/cache metering. Five issues found and fixed.
The dedup mechanism (request_id PK, proxy/session source isolation) is
untouched, so no double-counting is introduced.
- A (Claude + openai_chat, streaming): inject stream_options.include_usage
so OpenAI-compatible upstreams emit usage in the SSE tail. Without it the
converted Anthropic message_delta was all-zero and the whole request's
input/output/cache was dropped. Same root cause as the already-fixed
Codex Chat path; the injection is extracted into a shared helper
(transform::inject_openai_stream_include_usage) reused by both paths.
- C (Claude + gemini_native): subtract cachedContentTokenCount from
input_tokens in build_anthropic_usage so input becomes fresh input
(Anthropic semantics). Previously the cache-hit tokens were billed twice
because this path meters as app_type="claude" (input_includes_cache_read
= false) while Gemini's promptTokenCount includes the cache.
- D (Codex + openai_chat, streaming): gate log_usage on
has_billable_tokens() to skip the synthetic all-zero usage the converter
emits when a non-compliant upstream omits usage, preventing empty-row
request-count inflation.
- P2 (from_claude_stream_events): use has_billable_tokens() for the return
gate instead of input>0||output>0, so a fully-cached streamed request
(cache_read>0, input==output==0) is still recorded. Affects all
Claude-streaming paths, not just Gemini.
- P3 (Codex Chat->Responses, non-streaming): apply the same
has_billable_tokens() filter the streaming branch got, since the
synthesized all-zero usage makes from_codex_response return Some and
bypass the `if let Some` guard.
Add TokenUsage::has_billable_tokens() as the unified predicate. New tests
cover include_usage injection, gemini input subtraction, the gate itself,
cache-only stream recording, and synthetic all-zero codex usage.
Full lib suite: 1569 passed.
* fix(proxy): strip cache_control from OpenAI format conversion (#3805)
- Remove cache_control passthrough from system messages, text blocks,
and tools to prevent 400 errors on strict OpenAI-compatible endpoints
- Always simplify single text block content to plain string format
- Fixes two format conversion bugs reported in issue #3805
* fix(proxy): apply cargo fmt to fix CI formatting check
Convert Responses input_file (requiring file_id or file_data, never file_url which Chat file parts do not support) and input_audio parts into their Chat Completions equivalents, and handle top-level input_* items that previously fell through and were dropped, clearing stale pending reasoning for non-assistant messages.
Replace the unconditional finalize at chat-to-responses stream end with a three-way guard: complete normally when finish_reason or [DONE] arrived, emit an incomplete response when substantive output exists without a finish_reason, and emit a failed (stream_truncated) event for empty truncation instead of masking it as completed. Also propagate late-arriving reasoning_content onto still-active tool-call items.
Generalize the cross-turn reasoning cache in codex chat history from function_call only to the full tool-call triad (function_call, custom_tool_call, tool_search_call) and their *_output counterparts, so apply_patch and tool-search calls keep their reasoning_content when restored via previous_response_id.
Strict OpenAI-compatible upstreams (vLLM, enterprise gateways) reject
requests that carry tool_choice or parallel_tool_calls without a
non-empty tools array, returning 503/400 with:
"When using `tool_choice`, `tools` must be set."
The Responses→Chat Completions converter unconditionally forwarded
tool_choice and parallel_tool_calls even when tools ended up absent
or empty after conversion. This commit adds a guard that removes both
fields when the resulting tools array is missing or empty.
Added 9 regression tests covering:
- tool_choice dropped when tools absent
- tool_choice dropped when tools is empty array
- parallel_tool_calls dropped when tools absent
- tool_choice dropped when all tools filtered (e.g., missing name)
- tool_choice preserved when tools present (auto + function type)
- clean output when neither tool_choice nor tools present
- tool_choice 'none' dropped when no tools
- tool_search_output providing tools keeps tool_choice
Refs #3557
Co-authored-by: yueqi.guo <guo_yueqi@qq.com>
* fix(copilot): raise infinite-whitespace threshold from 20 to 500
The previous threshold of 20 falsely aborted legitimate tool calls whose
arguments contain indented code (write_file / edit_file with 4-8 levels
of indentation in Python / YAML / Rust / Markdown easily exceed 20
consecutive whitespace chars, especially when newlines are counted).
The real infinite-whitespace bug emits hundreds to thousands of
consecutive whitespace characters, so 500 keeps the safety net while
drastically reducing false positives.
Refs #2646
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* docs: clarify whitespace threshold comment
---------
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Co-authored-by: Jason <farion1231@gmail.com>
Custom (freeform) Codex tools are bridged through Chat Completions as
JSON `{"input": "..."}` functions, but the Chat->Responses stream still
re-emitted them via `response.function_call_arguments.*`, leaking the
JSON wrapper and using event types the Codex client does not route for
freeform tools.
Emit `response.custom_tool_call_input.delta`/`.done` (with the unwrapped
input text) for custom tool calls instead, suppressing the intermediate
function-argument deltas since a partial `{"input":` fragment cannot be
safely unwrapped mid-stream. Custom tool call items now use a `ctc_`
item-id prefix (matching the input events' item_id) consistently across
the streaming and non-streaming paths via a shared
response_tool_call_item_id_from_chat_name helper.
Covered by new streaming and non-streaming custom-tool tests.
When proxying Codex Responses requests through the Chat Completions
format (api_format=openai_chat), the request transform only forwarded
plain `function` tools and silently dropped `tool_search`, `namespace`
(MCP), and `custom` tools, so third-party APIs never saw the official
Codex plugins and could not call them.
Introduce CodexToolContext, built from the original Responses request,
which flattens all four tool kinds into Chat functions and keeps a
chat_name -> CodexToolSpec map. The response path (streaming and
non-streaming) looks names up in this map to restore the original
tool_search_call / custom_tool_call / namespaced function_call items
instead of reparsing the flattened name. Long namespaced names are
truncated with a sha256 suffix to fit the 64-char Chat tool-name limit.
Covered by new round-trip tests for all four tool kinds across both
streaming and non-streaming paths.
* fix(usage): resolve per-app credentials for native balance/coding-plan queries
The native usage-query paths (balance + coding_plan) in
`query_provider_usage_inner` read credentials only from `env.ANTHROPIC_*`.
That matches Claude providers, but Codex stores its key in
`auth.OPENAI_API_KEY` with the base URL inside a TOML `config` string,
Hermes/OpenClaw flatten them at the top level, and OpenCode nests them under
`options`. So the card "refresh usage" / auto-query returned empty
credentials and failed ("查询失败") for those apps, even though the
config-page "Test" button worked (the frontend extracts per-app correctly).
Introduce a single per-app resolver `Provider::resolve_usage_credentials`
that mirrors the frontend `getProviderCredentials`, and route both native
branches through it. Add `extract_codex_base_url` to `codex_config` as the
canonical Codex TOML base-URL parser and make the proxy adapter delegate to
it (removing a duplicate copy). Align the frontend `getProviderCredentials`
to cover OpenCode and Claude Desktop and to use the same OpenRouter/Google
key fallbacks as the backend.
Fixes#3158Fixes#3100Fixes#2625
* refactor(usage): explicit AppType arms + frontend trailing-slash trim
Address two review nits on the per-app credential resolver:
- provider.rs: replace the catch-all `_` arm in resolve_usage_credentials with an explicit `AppType::Claude | AppType::ClaudeDesktop` arm, so a new AppType variant fails to compile here instead of silently defaulting to the Anthropic env shape.
- UsageScriptModal.tsx: normalize getProviderCredentials baseUrl through a single trailing-slash trim so the frontend 'Test' path matches the backend resolver (trim_end_matches), keeping front/back truly in lockstep.
No behavior change for well-formed configs; tests/typecheck/fmt/clippy clean.
* fix(usage): skip empty primary credential fields in fallback chain
`obj.get(key)` returns `Some` for a present-but-empty field, so the
`.or_else()` fallback chains for the Claude/ClaudeDesktop and Gemini api-key
lookups only skipped *absent* keys, not empty ones. Presets seed fields like
`ANTHROPIC_AUTH_TOKEN` as present-but-empty placeholders, so a provider whose
real key lives in a fallback field (`ANTHROPIC_API_KEY` / `OPENROUTER_API_KEY`
/ `GOOGLE_API_KEY`, or `GOOGLE_API_KEY` for Gemini) resolved to an empty key on
the native balance/coding-plan path — while the frontend `a || b` (which skips
empty strings) still found it, reproducing the same Test-works / refresh-fails
divergence this PR removes.
Add a `first_non_empty` helper that skips present-but-empty values, matching
the frontend `||` semantics, and use it for both fallback chains. Tests cover
empty primary + populated fallback for Claude and Gemini.
* fix: add kimi/moonshot to Anthropic tool thinking history normalizer
Kimi's Anthropic-compatible endpoint requires reasoning_content in
assistant tool call messages when thinking is enabled, same as DeepSeek.
The is_reasoning_content_compatible_identifier already included kimi/moonshot
but is_anthropic_tool_thinking_history_identifier was missing them, causing
400 errors on multi-turn conversations with tool calls.
Closes#3351
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* refactor: unify reasoning vendor hints into a single SSOT list
Replace the two separately-maintained vendor identifier lists with one
shared REASONING_VENDOR_HINTS constant and a single
is_reasoning_vendor_identifier predicate. The Anthropic
tool-thinking-history matcher and the openai_chat reasoning_content
matcher previously kept independent lists, which is exactly how
kimi/moonshot ended up in one but not the other (#3351). A single source
of truth keeps the two from drifting apart again.
Also add a Kimi regression test for the Anthropic tool-history
normalization path.
---------
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-authored-by: Jason <farion1231@gmail.com>
Bump the default Opus route/model from claude-opus-4-7 to claude-opus-4-8
across provider presets (claude, claudeDesktop, hermes, openclaw, opencode,
universal), i18n locales (zh/en/ja/zh-TW), pricing seed data, and the
user-manual docs.
- Add claude-opus-4-8 pricing row ($5/$25/$0.50/$6.25); keep the 4-7 row
for historical usage stats (seeded via INSERT OR IGNORE).
- Claude Desktop proxy: accept bidirectional opus 4-7 <-> 4-8 route alias
during rollout so previously saved routes keep resolving.
- thinking_optimizer: route opus-4-8 through adaptive thinking and normalize
dotted model ids (also fixes dotted 4-6/4-7 falling back to legacy).
- usage_stats: normalize Bedrock/Vertex/aggregator opus-4-8 ids to base
pricing.
Also merge role:"system" messages into the Gemini systemInstruction in the
Anthropic->Gemini transform.
* feat: add MiMo reasoning_content support for Claude Code proxy
MiMo requires reasoning_content to be passed back in multi-turn
conversations with tool calls. Add "mimo" and "xiaomimimo" to the
preserve_reasoning_content whitelist so thinking blocks are converted
to reasoning_content when proxying Claude Code → MiMo.
Also handle redacted_thinking blocks (encrypted by Claude Code across
turns) by injecting a placeholder to prevent MiMo 400 errors.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix: support MiMo reasoning histories
- Fix the PR CI clippy failure for redacted thinking handling.
- Preserve MiMo reasoning content on the OpenAI Chat proxy path.
- Normalize Claude Desktop Anthropic thinking history for MiMo local routes.
- Reject unsupported Claude Desktop direct model remaps.
---------
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-authored-by: Jason <farion1231@gmail.com>
Codex provider switches now only write config.toml for third-party providers,
injecting the API key as experimental_bearer_token. The user's auth.json
(ChatGPT OAuth tokens) is preserved. Official providers with login material
still write auth.json normally. Backfill restores bearer tokens into stored
provider auth.OPENAI_API_KEY to maintain canonical shape.
Harden the proxy module against panics by removing optimistic
unwrap()/expect() calls in favor of pattern matching and graceful
fallbacks:
- copilot_optimizer/cache_injector: bind Value::Array/String directly
instead of is_array()+as_array().unwrap(); use is_none_or and in-place
string mutation
- hyper_client: gate the raw-write path with if-let + filter instead of
has_cases + unwrap()
- gemini_shadow: recover poisoned RwLock via into_inner() rather than
panicking, with warn logging
- streaming_codex_chat: replace expect("tool state exists") with
let-else (return/continue)
- merge_tool_results: early-return when messages is absent
- sse: fall back to from_utf8_lossy on the UTF-8 boundary slice
No behavior change on the happy path; all proxy tests pass.
No-argument tool calls produced empty argument strings that passed
through both the streaming response path and the request transform
verbatim. Strict OpenAI-compatible upstreams (e.g. Minimax) reject
`arguments: ""` with a 400 "invalid function arguments json string",
while lenient ones (OpenAI, Kimi) silently treat it as an empty object.
Add canonicalize_tool_arguments / canonicalize_tool_arguments_str helpers
that coerce empty/whitespace arguments to "{}" and apply them at the four
tool-call argument sites (streaming finalize + three request/response
transform paths). Leave function_call_output content untouched, where an
empty string is valid.
Auto-detect each Chat-routed Codex provider's reasoning interface from
its name, base URL, and model, then inject the matching thinking
parameter without manual configuration:
- Platform-first inference (OpenRouter, SiliconFlow) overrides model
rules, since the same model exposes different reasoning controls
depending on the hosting platform.
- Effort tiers are forwarded only to providers that support them
(DeepSeek, OpenRouter, and StepFun's step-3.5-flash-2603); on/off-only
providers (Kimi, GLM, Qwen, MiniMax, MiMo, SiliconFlow) drop the level
instead of sending a field the upstream rejects.
- OpenRouter uses the native reasoning:{effort} object, clamps max to
xhigh (its enum has no max), and forwards an explicit effort:"none" so
reasoning can be turned off.
- StepFun falls back to inference so per-model effort support is honored
(the static preset would have forced effort on step-3.5-flash too).
Includes the Codex provider-form reasoning controls, i18n strings
(zh/en/ja), and response-side reasoning extraction.
Responses-to-Chat conversion did not set stream_options.include_usage,
so OpenAI-compatible upstreams (kimi/MiniMax, etc.) omitted the trailing
usage chunk in streaming mode. This caused token/cost/cache stats to be
recorded as zero for third-party streaming requests routed through the
Codex Chat path.
Inject include_usage when the request is streaming, merging into any
client-provided stream_options instead of overwriting it. Add tests for
the streaming, non-streaming, and merge cases.
Thinking models like kimi/Moonshot and DeepSeek reject any assistant
message carrying tool_calls without a non-empty reasoning_content. When
cross-turn history recovery misses (proxy restart drops the in-memory
cache, ambiguous call_id, or a turn produced no reasoning upstream), the
converted assistant message has none and the whole request fails with
'reasoning_content is missing in assistant tool call message'.
Add a final-stage backfill that runs after all input items are
processed, so genuine trailing reasoning items still attach first and a
placeholder is only injected when none remains. Mirrors the placeholder
behavior in transform::anthropic_to_openai_with_reasoning_content.
The Codex Chat-to-Responses bridge already rewrites successful upstream
responses to the Responses shape, but the error branch in
handle_codex_chat_to_responses_transform passed Chat-shaped error bodies
through untouched (MiniMax base_resp, raw OpenAI Chat error, text/plain
"Unauthorized" pages, etc.), leaving Codex clients unable to recognize the
error.
Add chat_error_to_response_error in transform_codex_chat to regularize all
upstream error shapes into the standard {error: {message, type, code, param}}
envelope, then wire it through a new handle_codex_chat_error_response that
preserves the original HTTP status code. Non-JSON error bodies (HTML, plain
text) are wrapped as Value::String and truncated to 1KB at a UTF-8 char
boundary to keep diagnostic context without flooding the response.
Also fix a pre-existing append-vs-insert pitfall in three rebuilt-body
branches (Claude transform, Codex Chat normal, Codex Chat error): http
Builder::header is append, so leaking the upstream Content-Type produced
two Content-Type headers when the rewritten body was JSON. Remove the
upstream value before writing application/json.
Codex Chat providers (apiFormat=openai_chat, e.g. DeepSeek, MiniMax, Kimi)
were incorrectly probed via /v1/responses by Stream Check, which the upstream
rejects. Detect chat routing via codex_provider_uses_chat_completions and
issue the probe against /chat/completions with a Chat-shaped body.
Also align URL fallback order with the production CodexAdapter::build_url:
origin-only base URLs (https://api.deepseek.com) must hit /v1/<endpoint>
first; bare /<endpoint> only as a fallback. The previous order made any
non-404 error on the bare path (401/400/405) flag the provider as down even
though /v1/<endpoint> would have succeeded.
- Lift origin-only detection into proxy::providers::is_origin_only_url so
both build_url and Stream Check share a single source of truth.
- Collapse resolve_codex_stream_urls and resolve_codex_chat_stream_urls into
a generic resolve_codex_endpoint_urls(base_url, is_full_url, endpoint),
which also gives the Responses path a symmetric "full endpoint without
is_full_url flag" fallback for free.
- Restrict reasoning_effort propagation in the Chat branch to OpenAI o-series
models, mirroring transform_codex_chat's runtime check.
MiniMax's OpenAI-compatible chat endpoint strict-rejects any non-leading
role=system message with "invalid params, chat content has invalid message
role: system (2013)". The Codex client uses role=developer (and occasionally
role=system) to inject collaboration_mode / permissions / skills blocks
mid-conversation, and responses_role_to_chat_role maps both to chat's
system role. The converted messages array therefore frequently contained
system entries past index 0, which DeepSeek and OpenAI tolerate but MiniMax
flags as 2013.
Collapse all system messages into a single leading system message before
returning the chat request. The rewrite preserves every system fragment
(joined by "\n\n" in original order) and leaves non-system messages
untouched, so it is lossless for permissive backends as well.
- Add collapse_system_messages_to_head in transform_codex_chat.rs.
- Run it on the messages vector at the end of responses_to_chat_completions
before serializing.
- Cover the new path with two unit tests: one repros the MiniMax-shaped
input (developer items between users) and asserts no system role past
index 0; the other verifies non-system order is preserved and content
is joined with "\n\n".
When Codex client sends a model from the catalog (e.g., user selected via /model),
preserve that choice instead of always replacing with config.toml's default model.
- Check if request model exists in modelCatalog before falling back to config model
- Remove CODEX_CHAT_CLIENT_MODEL constant (no longer needed)
- Add test coverage for catalog model preservation
- Add Codex provider API format selection and model mapping for Chat-only upstreams.
- Convert Codex Responses requests to Chat Completions and rebuild Chat responses as Responses output.
- Preserve reasoning_content across non-streaming, streaming, tool calls, and previous_response_id follow-ups.
- Add a bounded Codex Chat history cache for restoring tool calls before tool outputs.
- Cover Chat bridge, compact routing, streaming, and history recovery with focused tests.
- Split leading <think> blocks from Chat content into Responses reasoning output.
- Keep assistant text output free of inline thinking tags.
- Support streamed inline thinking blocks before normal text deltas.
- Add regression coverage for MiniMax-style inline thinking responses.
- Add a Codex API format selector and routing badge for Chat Completions providers.
- Convert Codex Responses requests to upstream Chat Completions when routing is required.
- Convert Chat Completions JSON and SSE responses back to Responses format.
- Keep generated Codex wire_api values on Responses for Codex compatibility.
- Add i18n labels, provider metadata handling, and focused conversion tests.
* fix(gemini-native): resolve functionResponse.name and thought_signature replay for synthesized tool call IDs
Two related bugs in the Gemini Native format conversion layer:
1. **functionResponse.name resolution** (422 error): When Gemini's parallel
function calls omit the id field, cc-switch synthesizes gemini_synth_*
IDs. These are stored in the shadow store but can be lost in long sessions,
causing subsequent tool_result blocks to fail. Fix: pre-scan all assistant
messages in the request body to seed the tool_name_by_id map, and add a
last-resort fallback that scans the current content array for matching
tool_use blocks.
2. **thought_signature replay** (400 error): The Anthropic Messages format
strips thoughtSignature from tool_use blocks, but Gemini requires it on
every functionCall in multi-turn tool-use exchanges. Fix: build a
thought_signature_by_id map from shadow turns and attach thoughtSignature
when converting tool_use back to functionCall.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
* style: run cargo fmt on transform_gemini.rs
---------
Co-authored-by: Tiancrimson <tiancrimson@gmail.com>
Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
Co-authored-by: Jason <farion1231@gmail.com>
claude.rs and gemini.rs each defined an identical `hv` closure that wrapped
`HeaderValue::from_str` into a ProxyError::AuthError result, and codex.rs
spelled the same conversion out inline. /simplify reviewers flagged this
as drift-prone copy-paste.
Move the conversion into a single `pub fn auth_header_value` in
providers/adapter.rs and have the three adapters import it locally. Same
error wording everywhere, one place to update if HeaderValue semantics
ever change.
The Chat-Completions transformer used to forward tool_choice verbatim,
but the two APIs disagree on shape:
Anthropic "any" | {"type":"tool","name":"X"}
OpenAI Chat "required" | {"type":"function","function":{"name":"X"}}
Pass-through made the upstream return 400 for any tool-forcing client
(Claude Code, Copilot, etc.). The Responses-API transformer already had
the equivalent map_tool_choice_to_responses helper; this commit adds a
sibling map_tool_choice_to_chat with the chat-specific *nested* function
selector and five regression tests covering string / object × any /
auto / none / tool.
The two helpers are intentionally not merged: the difference between
flat and nested function selectors is exactly what the original bug
was, so keeping them as separate self-documenting functions reduces the
chance of the same regression returning.
User-pasted API keys can contain control chars or CR/LF that make
HeaderValue::from_str return Err; the previous unwrap inside every
adapter turned such input into a process-wide panic instead of a request
error. The trait now returns Result<_, ProxyError>; Claude/Codex/Gemini
impls propagate ProxyError::AuthError so the client sees a 401 with the
underlying parse error instead of a crash. Adds a regression test that
pastes a CRLF-containing key and asserts AuthError.
prompt_cache_key was falling back to provider.id when the client did not
supply a session, which collapsed every conversation onto a single key
and defeated upstream prefix caching. Only emit the key when a real
client-provided session/thread identity is available; otherwise let the
upstream use its default matching behaviour.
Additional fixes that affect cache stability:
- Canonicalise (sort) JSON keys in outgoing request bodies and in
tool_call arguments / tool_result content so semantically identical
requests produce identical byte sequences for upstream prefix caches.
- Exempt JSON Schema property maps (properties, patternProperties,
definitions, \$defs) from the underscore-prefix filter so user-defined
schema keys like _id and _meta survive.
- Add a [CacheTrace] debug log with stable hashes for instructions,
tools, input and include to help diagnose cache misses.
- Thread session_id into the usage logger for request correlation.
Komi