Use activation mode#

By default, conda-exec only prepends the environment’s bin/ directory to PATH. The --activate flag asks conda’s activator to compute the environment variables for the subprocess instead.

What –activate does#

With --activate, conda-exec runs conda’s activator before executing the tool. This:

Sets CONDA_PREFIX to the ephemeral environment path
Exports environment variables reported by conda’s activator
Unsets variables that activation marks for removal

Without --activate, only PATH is modified.

conda-exec does not execute package activate.d/ shell scripts. If a tool requires those scripts, use conda create or conda env create for a named environment and run it with conda run or an explicitly activated shell. That path matches conda’s full shell activation semantics.

When you need activation#

Some tools check CONDA_PREFIX, CONDA_DEFAULT_ENV, or other variables reported by conda’s activator. Common cases:

Bioinformatics pipelines that check CONDA_PREFIX to locate data files
Tools that use the active environment name for configuration
Scripts that shell out to conda-aware tools

conda exec --activate -c bioconda snakemake --cores 4

When you do not need activation#

Tip

Most CLI tools work without activation. If you are running linters, formatters, or similar standalone tools, skip --activate for faster execution.

Most standalone CLI tools only need the binary on PATH and work fine without activation. For example, linters, formatters, and build tools typically do not inspect CONDA_PREFIX:

conda exec ruff check .
conda exec black --check .
conda exec jq '.name' package.json

Skipping activation is the default because it avoids the activation calculation on every run.

Performance#

Activation adds overhead to each invocation because it imports conda’s activation machinery and computes the activated environment. For tools that do not need it, omitting --activate avoids this cost.

Example with a script#

Activation also works in script mode:

conda exec --activate script.py

The script runs with activation environment variables such as CONDA_PREFIX set, which is useful for scripts that call conda-aware libraries.