Manage Python environments, use Pyright and Ruff, inspect models, generate type stubs, run Pytest, and profile supported workloads in Flame IDE.
When two services use different Python environments, a passing test in the wrong interpreter tells you little. Flame keeps environment selection per project, brings Pytest results into the test explorer, and provides tools for type stubs and model inspection. Use the source, terminal, API client, and Git diff to verify the complete change.
Discover virtualenv, Conda, Poetry, Pipenv, system, and custom environments. Select the executable per project, create a virtualenv through uv or venv, and use supported isolated runtime downloads when appropriate.
Generate .pyi stubs for supported targets and inspect Pydantic, dataclass, NamedTuple, and TypedDict models. Read the generated schema and validate a sample JSON payload before changing an API consumer.
Run Pytest, inspect parameterized cases and fixture relationships, and view coverage when the project supplies it. Python CPU profiling uses py-spy with a cProfile fallback; memory profiling depends on the supported tools available on the host.
Open the Python service and select its intended interpreter. Create or discover the environment, then install the repository's dependencies using its own package and lockfile workflow.
Run the relevant Pytest case in the test explorer or project terminal. Confirm the interpreter and inputs before asking an agent to investigate.
Open the model inspector for the affected source, examine fields and schema, and validate a representative JSON payload. Generate a type stub when an untyped boundary needs one, then review its signatures.
Make a focused correction, or attach the failing test and model context to an agent. Rerun the regression and normal-case tests, inspect the Git diff, and verify the endpoint if its behavior changed.