Troubleshoot Conda Code#

Two conda branches appear#

Python Environments registers its built-in ms-python.python:conda manager independently. Its public API does not currently let another extension replace or hide that manager.

Select jezdez.conda-code:conda as the default and project manager. See Select the Conda Code provider.

The Conda Code branch is empty#

Open View: Output and select Conda Code.
Run Conda Code: Refresh Environments.
Check the configured executable with conda info --json.
Set conda-code.condaExecutable to an absolute path if needed.

Conda Code retains environments without Python, but marks them with a warning. A prefix without a conda-meta directory is not a conda environment and is ignored.

The configured executable selects the primary conda. Additional conda installations can be discovered through ~/.conda/environments.txt, conda executables located inside their roots, relevant conda environment variables, and standard installation locations.

Initial regular discovery does not wait for conda info --json. Conda Code loads missing or stale conda details in the background and caches the result. It watches the exact configuration files reported by conda while VS Code is open and fingerprints those files, relevant conda environment inputs, and the configured executable across restarts. Run Conda Code: Refresh Environments to force that request after changing a new or unreported configuration source. The 24-hour cache age is the final fallback where no source or file event was available.

An environment cannot be changed#

Conda Code keeps a valid prefix visible when it cannot determine an owning conda executable. Selection, package listing, and inspection remain available. Direct execution remains available when the prefix contains Python. Shell activation, package changes, and deletion require an owning conda executable. Make conda available in the owning installation, then refresh.

A workspace does not appear#

Check all of the following:

The manifest directory is registered as a Python project.
conda workspace info --json succeeds for the manifest.
The environment is installed.
The manifest is not a Pixi workspace ignored because Pixi Code is installed.
No other workspace manifest reports the same prefix.

Only installed workspace environments appear. Use Python Envs: Create Environment to install a declared environment.

A workspace prefix disappears#

When two workspace manifests report the same prefix, Conda Code hides that prefix and refuses changes. Change one manifest so that each prefix has one reporting workspace, then refresh.

Package changes are refused#

Conda Code enables workspace changes only when conda-workspaces reports enough metadata to preserve the manifest:

Adding to a composed or zero-feature environment needs a complete snapshot.
Removal and update need the direct dependency’s declaration location.
PyPI dependencies can be removed, but not updated.
Transitive dependencies remain read-only.
A PyPI package remains read-only when conda-pypi gave its installed conda record a different name from the manifest declaration.

The 0.7-compatible path can still add to an environment backed by exactly one feature. See Use conda-workspaces.

SBOM export is refused or the format is missing#

Make sure the active editor scope selects an environment from the jezdez.conda-code:conda manager. A regular environment also needs a usable owner executable. Refresh Conda Code after correcting its conda installation.

Install conda-sboms 0.3.0 or newer beside the executable that Conda Code routes for the selected environment, then confirm that the same executable lists cyclonedx-json-v1.7 in its export --help output. Regular environments use their owning conda installation. Workspace environments use the configured primary conda installation.

Quick Create reports multiple project inputs#

The project root contains more than one recognized environment input. Run Python Envs: Create Environment and choose one, or remove the inputs that do not apply.

An existing discovered workspace takes precedence and does not use these files.

Additional creation packages are refused#

explicit.txt, conda-lock.yml, and conda-lock.yaml describe exact package sets. Conda Code refuses additional creation packages and disables configured default packages for these inputs.

Add packages after creation only when the environment should diverge from the exact input.

A conda-global environment is missing#

This is expected. Conda Code excludes the active conda-global tool root because those prefixes back globally installed commands rather than selectable project environments.