Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.
Python agentaianthropicclaude-codecompressioncontext-engineeringcontext-windowcursorfastapilangchainllmmcpopenaiprompt-engineeringproxypythonragtoken-optimizationtokenstypescript
fix(codex): route alpha search through the Codex backend (#2538)
Rod Boeva540eb2c61b1
diff --git a/headroom/providers/proxy_routes.py b/headroom/providers/proxy_routes.pyindex b7c6c595b..e77838bb0 100644--- a/headroom/providers/proxy_routes.py+++ b/headroom/providers/proxy_routes.py@@ -7,13 +7,17 @@ import loggingfrom typing import Anyfrom fastapi import FastAPI, Request, WebSocket+from fastapi.responses import Responsefrom headroom.providers.cloudcode import normalize_cloudcode_passthrough_path+from headroom.providers.codex.endpoints import codex_backend_url+from headroom.providers.codex.headers import drop_headerfrom headroom.providers.codex.live import (CODEX_LIVE_ROUTE_PATHS,handle_codex_live_websocket,)from headroom.providers.codex.responses import handle_chatgpt_codex_responses_subpath+from headroom.providers.codex.runtime import resolve_codex_routingfrom headroom.providers.model_metadata import (MODEL_METADATA_LIST_ENDPOINT,handle_model_metadata_endpoint,@@ -67,6 +71,32 @@ from headroom.proxy.request_scope import normalize_request_pathlogger = logging.getLogger("headroom.proxy.routes")+async def _handle_chatgpt_codex_alpha_search(request: Request, proxy: Any) -> Response | None:+ upstream_headers = dict(request.headers.items())+ drop_header(upstream_headers, "host")+ drop_header(upstream_headers, "accept-encoding")+ from headroom.proxy.helpers import _strip_internal_headers++ decision = resolve_codex_routing(_strip_internal_headers(upstream_headers))+ if not decision.is_chatgpt_auth:+ return None++ body = await request.body()+ assert proxy.http_client is not None+ resp = await proxy.http_client.request(+ request.method,+ codex_backend_url("/alpha/search", request.url.query),+ headers=decision.headers,+ content=body,+ timeout=120.0,+ )+ return Response(+ content=resp.content,+ status_code=resp.status_code,+ headers=dict(resp.headers),+ )++def _register_provider_passthrough_route(app: FastAPI,proxy: Any,@@ -456,6 +486,16 @@ def register_provider_routes(app: FastAPI, proxy: Any) -> None:provider_name=provider_name,)+ @app.post("/v1/alpha/search")+ async def codex_alpha_search(request: Request):+ chatgpt_response = await _handle_chatgpt_codex_alpha_search(request, proxy)+ if chatgpt_response is not None:+ return chatgpt_response+ return await proxy.handle_passthrough(+ request,+ _select_passthrough_base_url(proxy, dict(request.headers)),+ )+_register_openai_image_routes(app, proxy)_register_codex_live_routes(app, proxy)diff --git a/tests/test_provider_proxy_routes.py b/tests/test_provider_proxy_routes.pyindex fe7f9ddd0..2319e3571 100644--- a/tests/test_provider_proxy_routes.py+++ b/tests/test_provider_proxy_routes.py@@ -274,6 +274,133 @@ def test_provider_passthrough_routes_forward_expected_targets(monkeypatch) -> Noassert len(anthropic_calls) >= 2+def test_codex_alpha_search_route_from_headroom_issue_2525() -> None:+ class FakeAsyncClient:+ def __init__(self) -> None:+ self.calls: list[tuple[str, str, dict[str, str], bytes]] = []++ async def request(self, method, url, **kwargs): # type: ignore[no-untyped-def]+ self.calls.append(+ (+ method,+ url,+ dict(kwargs.get("headers", {})),+ kwargs.get("content", b""),+ )+ )+ return httpx.Response(200, json={"ok": True})++ async def aclose(self) -> None:+ return None++ with TestClient(_app()) as client:+ fake_http_client = FakeAsyncClient()+ client.app.state.proxy.http_client = fake_http_client+ response = client.post(+ "/v1/alpha/search?query=weather",+ headers={+ "Authorization": "Bearer oauth-token",+ "ChatGPT-Account-ID": "acct_123",+ },+ json={"query": "weather"},+ )++ assert response.status_code == 200+ assert response.json() == {"ok": True}+ assert len(fake_http_client.calls) == 1+ method, url, headers, body = fake_http_client.calls[0]+ assert method == "POST"+ assert url == "https://chatgpt.com/backend-api/codex/alpha/search?query=weather"+ assert headers["authorization"] == "Bearer oauth-token"+ assert headers["chatgpt-account-id"] == "acct_123"+ assert headers["content-length"] == "19"+ assert headers["content-type"] == "application/json"+ assert body == b'{"query":"weather"}'+++def test_non_chatgpt_alpha_search_falls_through_to_openai_upstream(monkeypatch) -> None:+ calls: list[tuple[str, str, str, str, str]] = []++ async def fake_passthrough(self, request, base_url, sub_path="", provider_name=""): # type: ignore[no-untyped-def]+ calls.append((request.method, request.url.path, base_url, sub_path, provider_name))+ return JSONResponse(+ {+ "base_url": base_url,+ "sub_path": sub_path,+ "provider": provider_name,+ }+ )++ monkeypatch.setattr(HeadroomProxy, "handle_passthrough", fake_passthrough)++ with TestClient(_app()) as client:+ response = client.post(+ "/v1/alpha/search",+ headers={"Authorization": "Bearer sk-proj-openai-test"},+ json={"query": "weather"},+ )++ assert response.status_code == 200+ assert response.json() == {+ "base_url": "https://api.openai.test",+ "sub_path": "",+ "provider": "",+ }+ assert calls == [+ (+ "POST",+ "/v1/alpha/search",+ "https://api.openai.test",+ "",+ "",+ )+ ]+++def test_codex_alpha_search_route_matrix(monkeypatch) -> None:+ fallback_calls: list[tuple[str, str]] = []++ async def fake_passthrough(self, request, base_url, sub_path="", provider_name=""): # type: ignore[no-untyped-def]+ fallback_calls.append((request.url.path, base_url))+ return JSONResponse({"base_url": base_url, "provider": provider_name})++ monkeypatch.setattr(HeadroomProxy, "handle_passthrough", fake_passthrough)++ class FakeAsyncClient:+ def __init__(self) -> None:+ self.urls: list[str] = []++ async def request(self, method, url, **kwargs): # type: ignore[no-untyped-def]+ self.urls.append(url)+ return httpx.Response(200, json={"ok": True})++ async def aclose(self) -> None:+ return None++ with TestClient(_app()) as client:+ fake_http_client = FakeAsyncClient()+ client.app.state.proxy.http_client = fake_http_client++ oauth_response = client.post(+ "/v1/alpha/search",+ headers={+ "Authorization": "Bearer oauth-token",+ "ChatGPT-Account-ID": "acct_123",+ },+ json={"query": "weather"},+ )+ api_key_response = client.post(+ "/v1/alpha/search",+ headers={"Authorization": "Bearer sk-proj-openai-test"},+ json={"query": "weather"},+ )++ assert oauth_response.status_code == 200+ assert api_key_response.status_code == 200+ assert fake_http_client.urls == ["https://chatgpt.com/backend-api/codex/alpha/search"]+ assert fallback_calls == [("/v1/alpha/search", "https://api.openai.test")]++def test_proxy_route_helpers_prefer_legacy_targets_and_gemini_passthrough() -> None:proxy_routes = importlib.import_module("headroom.providers.proxy_routes")proxy = type(