Command reference#
Use conda feature COMMAND --help for the installed parser’s exact options. Run the conda installation where the plugin is installed. If the bare command selects another installation, use that host interpreter’s python -m conda. See installation.
Queries#
list#
conda feature list [SCOPE] [--json]
Lists installed feature declarations and their state. It does not fetch a remote catalog. An empty catalog is valid. Human output shows feature ID, maturity, requested state, effective setting, readiness, and description.
explain FEATURE#
conda feature explain FEATURE [SCOPE] [--json]
Reports the provider identity, settings sources, readiness reasons, configurable settings, and guidance. It includes replacement/graduation metadata when present. The sources mapping identifies contributing files or environment variables without copying all their values.
diagnose#
conda feature diagnose [SCOPE] [--json]
Inspects selected features and deployment receipts adjacent to the chosen configuration file. Returns path, scope, problems, and a message explaining the limits of local checks. Each problem has feature, kind, message, and action. Diagnostic kinds and recovery actions are listed in the recovery guide.
Mutations#
enable FEATURE#
conda feature enable FEATURE [--with-dependencies] [SCOPE] [--dry-run | --yes]
Plans an opt-in or deployment. Checks compatibility, prerequisites, conflicts, scope, and prospective configuration. --with-dependencies includes supported prerequisite opt-ins at the selected scope. Package/custom prerequisites must be deployed separately. No implicit cross-scope writes occur.
config FEATURE#
conda feature config FEATURE --set KEY VALUE [--set KEY VALUE ...] [SCOPE] [--dry-run | --yes]
conda feature config FEATURE --set plugins.limit 4 [SCOPE] [--dry-run | --yes]
conda feature config FEATURE --get [KEY ...] [SCOPE]
conda feature config FEATURE --show [KEY ...] [SCOPE]
conda feature config FEATURE --show-sources [SCOPE]
conda feature config FEATURE --describe [KEY ...] [SCOPE]
conda feature config FEATURE --validate [SCOPE]
conda feature config FEATURE --remove-key KEY [--remove-key KEY ...] [SCOPE] [--dry-run | --yes]
Changes declared non-secret scalar settings. Repeat --set for multiple key/value pairs, using each key only once. The example key is illustrative. Conda normalizes values and validates their combined configuration. For step-backed deployments, disable before changing settings when an existing receipt is present.
Choose one action per invocation. All named keys must be in the feature’s declared non-secret settings. Use enable and disable to change its enabled flag.
Action |
Behavior |
|---|---|
|
Read normalized values saved in the selected file. Omit keys to read all declared settings that are saved there. |
|
Read effective values, including defaults and higher-priority overrides. Omit keys to show all declared settings. |
|
Show contributing values grouped by source, filtered to the feature’s declared settings. |
|
Show conda’s parameter descriptions, types, and defaults. Omit keys for all declared settings. |
|
Validate the combined conda configuration and local feature requirements without writing or solving packages. Unmet requirements produce an error. |
|
Remove the saved canonical key and any saved aliases, revealing inherited values or defaults. Repeat for several keys. Preview and confirmation are required as for |
Removal fails if the key is not saved, would break an enabled feature’s required settings, or belongs to an existing deployment that must be disabled first. Queries do not create files or receipts. JSON queries return values, sources, descriptions, or valid, alongside operation and feature details. A validation success does not verify runtime execution.
Conda validation errors omit configuration values from command output because those errors can include private, unrelated settings. Inspect invalid configuration locally. --show-sources validates and returns only declared settings from each source.
Sequence operations (--add, --prepend, --append, and --remove) are not supported because feature configuration currently declares scalar settings only.
The config command was added in 0.2.0. Release 0.1.0 supports setting values through the following syntax, which remains available:
conda feature configure FEATURE KEY=VALUE [KEY=VALUE ...] [SCOPE] [--dry-run | --yes]
disable FEATURE and disable --all#
conda feature disable FEATURE [SCOPE] [--dry-run | --yes]
conda feature disable --all [SCOPE] [--dry-run | --yes]
Disables one feature or all boolean-backed features at the selected scope. A feature name and --all cannot be combined. Individual disable protects enabled dependents. Bulk disable processes dependents first and identifies deployments that require separate commands. Neither form is an all-plugins emergency switch. Unconditional graduated features are outside disable controls.
repair FEATURE#
conda feature repair FEATURE [SCOPE] [--dry-run | --yes]
Retries an enabled choice saved at the selected scope using the normal enable plan. It preserves user edits and ownership checks and still requires readiness checks to pass. A fresh opt-in without a saved choice must use enable.
migrate FEATURE#
conda feature migrate FEATURE [SCOPE] [--dry-run | --yes]
Migrates a saved enabled boolean feature to its installed, provider-declared replacement. Both must support the selected scope. Transfers only mapped saved settings and refuses conflicting mapped destination settings. The replacement’s enabled flag is explicitly set to true, including when it was saved as false. Generic deployment migrations are not supported.
cleanup FEATURE#
conda feature cleanup FEATURE [SCOPE] [--dry-run | --yes]
Removes saved obsolete boolean opt-ins for features declared unconditionally graduated. It does not remove functionality, packages, or shell resources. The provider must retain the old setting declaration to identify it and its aliases.
Results and exit statuses#
The development version uses Rich tables and the conda-workspaces status palette: blue headings, cyan completed or ready states, yellow previews and warnings, and red errors. Labels carry the meaning without color. Narrow terminals use per-feature records rather than truncating names and states. Human-readable paths inside your home directory use ~/, while JSON retains exact paths. Redirected output has no color escapes by default, and Rich respects NO_COLOR. Dynamic provider text is literal, with terminal control characters shown as escape text. JSON output bypasses Rich and keeps its existing structure.
Mutation results include the operation, selected feature IDs, destination path, scope, public changes, and a message. Depending on the operation, they can also include removal keys, post-operation status, checked environments, and the scope of discovery. Private solved transactions, fingerprints, and recovery data are not returned as public plans.
Successful queries and previews exit 0. Handled operation errors exit 1. Argument parsing errors exit 2. An empty diagnostic problem list is not required for a successful diagnostic query. Unexpected runtime failures or conda-level failures may be handled by conda itself.
Noninteractive writes require --yes. JSON output is noninteractive even when stdin is a terminal. A no-op or dry-run does not write. Prompts and diagnostics go to stderr.
Conda doctor#
conda doctor --list
conda doctor features --prefix /path/to/environment
conda doctor features --fix --prefix /path/to/environment
The plugin registers the features health check and fixer. Conda owns doctor’s parser, text report, and confirmation behavior. For one complete JSON feature report, use conda feature diagnose --json instead.