Use Claude Code, Codex, VSCode, Pi, and OpenCode (and 6 other harnesses) for free (1.3B+ free tokens) from your terminal, app, IDE, or phone, and now from the browser with native browser sessions (multi-harness + multi-model) like OpenClaw (voice supported + ToS friendly)
Python
from copy import deepcopy
from typing import Any
from free_claude_code.core.anthropic import ReasoningReplayMode
from free_claude_code.core.anthropic.models import MessagesRequest
from free_claude_code.core.reasoning import ReasoningPolicy
from free_claude_code.providers.openai_chat.request_policy import (
OpenAIChatRequestPolicy,
build_openai_chat_request_body,
)
_POLICY = OpenAIChatRequestPolicy(
provider_name="TEST",
reasoning_replay=ReasoningReplayMode.THINK_TAGS,
)
def _request(name: str, *, continued: bool = False) -> MessagesRequest:
messages: list[dict] = [
{"role": "user", "content": "Use the tool"},
{
"role": "assistant",
"content": [
{
"type": "tool_use",
"id": "call_1",
"name": name,
"input": {"value": "x"},
}
],
},
{
"role": "user",
"content": [
{
"type": "tool_result",
"tool_use_id": "call_1",
"content": "done",
}
],
},
]
if continued:
messages.append({"role": "user", "content": "Continue"})
return MessagesRequest.model_validate(
{
"model": "test-model",
"messages": messages,
"tools": [
{
"name": name,
"description": "Long tool",
"input_schema": {
"type": "object",
"properties": {"value": {"type": "string"}},
},
},
{
"name": "safe_tool",
"description": "Safe tool",
"input_schema": {"type": "object"},
},
],
"tool_choice": {"type": "tool", "name": name},
}
)
def _body(request: MessagesRequest, *, postprocessors=()) -> dict[str, Any]:
return build_openai_chat_request_body(
request,
reasoning=ReasoningPolicy.provider_default(),
policy=_POLICY,
postprocessors=postprocessors,
)
def test_chat_request_uses_one_alias_in_definition_choice_and_history() -> None:
original = "mcp__long_server__" + "x" * 90
request = _request(original)
snapshot = request.model_dump()
body = _body(request)
tools = body["tools"]
assert isinstance(tools, list)
alias = tools[0]["function"]["name"]
assert alias != original
assert len(alias) <= 64
assert tools[1]["function"]["name"] == "safe_tool"
assert body["tool_choice"] == {
"type": "function",
"function": {"name": alias},
}
messages = body["messages"]
assert isinstance(messages, list)
assistant = next(message for message in messages if message["role"] == "assistant")
assert assistant["tool_calls"][0]["function"] == {
"name": alias,
"arguments": '{"value": "x"}',
}
assert assistant["tool_calls"][0]["id"] == "call_1"
assert request.model_dump() == snapshot
def test_chat_name_encoding_runs_after_provider_postprocessors() -> None:
original = "mcp__postprocessor_order__" + "x" * 80
request = _request(original)
observed: list[str] = []
def observe_names(body, _request, _reasoning) -> None:
observed.append(body["tools"][0]["function"]["name"])
body = _body(request, postprocessors=(observe_names,))
assert observed == [original]
assert body["tools"][0]["function"]["name"] != original
def test_chat_aliases_remain_stable_for_append_only_requests() -> None:
original = "mcp__cache_stable__" + "x" * 90
prefix_body = _body(_request(original))
continued_body = _body(_request(original, continued=True))
assert continued_body["tools"] == prefix_body["tools"]
assert continued_body["tool_choice"] == prefix_body["tool_choice"]
prefix_messages = prefix_body["messages"]
assert continued_body["messages"][: len(prefix_messages)] == prefix_messages
def test_chat_encoding_does_not_rewrite_unrelated_name_fields() -> None:
original = "mcp__canonical_name_only__" + "x" * 80
request = _request(original)
request.extra_body = {"metadata": {"name": original}}
policy = OpenAIChatRequestPolicy(
provider_name="TEST",
reasoning_replay=ReasoningReplayMode.THINK_TAGS,
include_extra_body=True,
)
snapshot = deepcopy(request.extra_body)
body = build_openai_chat_request_body(
request,
reasoning=ReasoningPolicy.provider_default(),
policy=policy,
)
assert body["extra_body"] == snapshot
assert body["tools"][0]["function"]["name"] != original