Skip to content

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

Fix Muse model catalog visibility and metadata (#1674)

Ali Khokhar8fb387011348
diff --git a/pyproject.toml b/pyproject.toml
index 32b6447e..3b85a2fa 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
[project]
name = "free-claude-code"
-version = "5.22.2"
+version = "5.22.3"
description = "Local proxy connecting coding agents to OpenAI-compatible AI providers"
readme = "README.md"
requires-python = ">=3.14.0"
diff --git a/smoke/features.py b/smoke/features.py
index 9bf99a45..ab04e796 100644
--- a/smoke/features.py
+++ b/smoke/features.py
@@ -205,7 +205,7 @@ FEATURE_INVENTORY: tuple[FeatureCoverage, ...] = (
"tests/cli/test_muse_launcher.py",
"tests/api/test_model_listing.py",
),
- ("test_muse_model_catalog_is_the_fixed_responses_projection",),
+ ("test_muse_model_catalog_marks_every_routable_model_visible",),
("test_muse_cli_headless_e2e",),
("clients",),
("Muse Code 0.2.1+",),
diff --git a/smoke/product/test_client_product_live.py b/smoke/product/test_client_product_live.py
index 5a2d523e..9accae20 100644
--- a/smoke/product/test_client_product_live.py
+++ b/smoke/product/test_client_product_live.py
@@ -823,6 +823,19 @@ def test_muse_cli_headless_e2e(smoke_config: SmokeConfig, tmp_path: Path) -> Non
assert auth_token not in combined
assert "GET /v1/models?view=responses" in server_log
assert "GET /muse-code/models" in server_log
+ # Muse accepts a generic OpenAI list but hides rows without its native
+ # metadata. Inspect the actual CLI's parsed catalog, not just HTTP success.
+ catalog_files = list(
+ (isolated_home / "data" / "muse" / "model-catalog").glob("*.json")
+ )
+ assert catalog_files, "Muse did not persist its parsed provider catalog"
+ visible_models = {
+ row["model_id"]
+ for catalog_file in catalog_files
+ for row in json.loads(catalog_file.read_text(encoding="utf-8"))["rows"]
+ if row["visibility"] == "visible"
+ }
+ assert full_model in visible_models, "Muse hides the configured FCC model"
responses_count = server_log.count("POST /v1/responses")
assert responses_count >= 1
assert responses_count == len(provider_requests)
diff --git a/src/free_claude_code/api/model_catalog.py b/src/free_claude_code/api/model_catalog.py
index 85f4cd5f..5dfd9edc 100644
--- a/src/free_claude_code/api/model_catalog.py
+++ b/src/free_claude_code/api/model_catalog.py
@@ -29,6 +29,24 @@ class ModelCatalogView(StrEnum):
RESPONSES = "responses"
+class MuseModelLimits(BaseModel):
+ context: int | None = None
+ output: int | None = None
+
+
+class MuseModelMetadata(BaseModel):
+ """Muse's native model visibility and capability metadata."""
+
+ name: str
+ is_hidden: Literal[False] = False
+ reasoning: bool | None = None
+ limit: MuseModelLimits | None = None
+
+
+class MuseMetadataEnvelope(BaseModel):
+ muse_code: MuseModelMetadata = Field(serialization_alias="muse-code")
+
+
class ModelResponse(BaseModel):
object: Literal["model"] = "model"
created: int = 0
@@ -65,6 +83,7 @@ class ModelResponse(BaseModel):
inference_idle_timeout_seconds: int | None = Field(
default=None, serialization_alias="inferenceIdleTimeoutSecs"
)
+ metadata: MuseMetadataEnvelope | None = None
class ModelsListResponse(BaseModel):
@@ -140,6 +159,33 @@ def build_models_list_response(
return _build_direct_models_response(settings, runtime, view=view)
+def build_muse_models_list_response(
+ settings: Settings, runtime: RequestRuntimePort
+) -> ModelsListResponse:
+ """Keep routable IDs while making them visible to Muse's native picker."""
+ catalog = build_models_list_response(
+ settings, runtime, view=ModelCatalogView.RESPONSES
+ )
+ for model in catalog.data:
+ limits = (
+ MuseModelLimits(
+ context=model.context_window_tokens,
+ output=model.max_output_tokens,
+ )
+ if model.context_window_tokens is not None
+ or model.max_output_tokens is not None
+ else None
+ )
+ model.metadata = MuseMetadataEnvelope(
+ muse_code=MuseModelMetadata(
+ name=model.display_name,
+ reasoning=model.supports_reasoning,
+ limit=limits,
+ )
+ )
+ return catalog
+
+
def _build_claude_models_response(
settings: Settings, runtime: RequestRuntimePort
) -> ModelsListResponse:
diff --git a/src/free_claude_code/api/routes.py b/src/free_claude_code/api/routes.py
index 3fc87d9c..0a3129d7 100644
--- a/src/free_claude_code/api/routes.py
+++ b/src/free_claude_code/api/routes.py
@@ -27,6 +27,7 @@ from .model_catalog import (
ModelCatalogView,
ModelsListResponse,
build_models_list_response,
+ build_muse_models_list_response,
)
from .ports import ApiServices
from .request_errors import ordinary_application_error_response
@@ -219,11 +220,7 @@ async def list_muse_models(
):
"""List the direct Responses models expected by Muse Code."""
trace_event(stage="ingress", event="free_claude_code.api.models.list", source="api")
- return build_models_list_response(
- settings,
- services.requests,
- view=ModelCatalogView.RESPONSES,
- )
+ return build_muse_models_list_response(settings, services.requests)
@router.post("/stop")
diff --git a/tests/api/test_model_listing.py b/tests/api/test_model_listing.py
index 6e7c56d7..b86bf670 100644
--- a/tests/api/test_model_listing.py
+++ b/tests/api/test_model_listing.py
@@ -309,7 +309,7 @@ def test_claude_model_view_does_not_expose_capability_fields():
)
-def test_muse_model_catalog_is_the_fixed_responses_projection():
+def test_muse_model_catalog_marks_every_routable_model_visible():
app = create_test_app(_settings(model_opus=None, model_haiku=None))
manager = provider_manager_for_app(app)
manager.cache_model_infos(
@@ -326,7 +326,14 @@ def test_muse_model_catalog_is_the_fixed_responses_projection():
assert responses.status_code == 200
assert muse.status_code == 200
- assert muse.json() == responses.json()
+ assert all("metadata" not in row for row in responses.json()["data"])
+ for row in muse.json()["data"]:
+ metadata = row["metadata"]["muse-code"]
+ assert metadata["is_hidden"] is False
+ assert metadata["name"] == row["display_name"]
+ assert "limit" not in metadata
+ original = {key: value for key, value in row.items() if key != "metadata"}
+ assert original in responses.json()["data"]
assert [row["id"] for row in muse.json()["data"]] == [
"deepseek/deepseek-chat",
"claude-3-freecc-no-thinking/open_router/plain-model",
@@ -334,6 +341,47 @@ def test_muse_model_catalog_is_the_fixed_responses_projection():
]
+def test_muse_catalog_uses_native_metadata_for_known_limits_and_reasoning():
+ app = create_test_app(_settings(model_opus=None, model_haiku=None))
+ provider_manager_for_app(app).cache_model_infos(
+ "open_router",
+ {
+ ProviderModelInfo(
+ "reasoning-model",
+ supports_thinking=True,
+ context_window_tokens=131072,
+ max_output_tokens=8192,
+ ),
+ ProviderModelInfo(
+ "plain-model",
+ supports_thinking=False,
+ context_window_tokens=65536,
+ ),
+ },
+ )
+ rows = {
+ row["provider_model_ref"]: row["metadata"]["muse-code"]
+ for row in TestClient(app).get("/muse-code/models").json()["data"]
+ }
+
+ assert rows["open_router/reasoning-model"] == {
+ "name": "open_router/reasoning-model",
+ "is_hidden": False,
+ "reasoning": True,
+ "limit": {"context": 131072, "output": 8192},
+ }
+ assert rows["open_router/plain-model"] == {
+ "name": "open_router/plain-model (no thinking)",
+ "is_hidden": False,
+ "reasoning": False,
+ "limit": {"context": 65536},
+ }
+ assert rows["deepseek/deepseek-chat"] == {
+ "name": "deepseek/deepseek-chat",
+ "is_hidden": False,
+ }
+
+
def test_responses_model_view_rounds_timeout_up_before_margin():
app = create_test_app(
_settings().model_copy(update={"provider_progress_timeout": 1.1})
diff --git a/uv.lock b/uv.lock
index 420d369e..c4b97471 100644
--- a/uv.lock
+++ b/uv.lock
@@ -598,7 +598,7 @@ wheels = [
[[package]]
name = "free-claude-code"
-version = "5.22.2"
+version = "5.22.3"
source = { editable = "." }
dependencies = [
{ name = "aiohttp" },