A tox environment may have git+ssh dependencies or even direct ssh command calls. In non-interactive shell sessions (such as in CI tasks) the identity is often available only through ssh-agent to keep the private key in secret or because there is no way to enter the passphrase. SSH client is communicating with ssh-agent session through the socket opened on SSH_AUTH_SOCK. This variable must be passed in the tox environment to SSH client could work with the agent session from there. If there is an active agent in a shell session, it most likely implies that it will be used for every SSH client connection even in tox environments.

ssh-agent evaluation exports not only SSH_AUTH_SOCK but also SSH_AGENT_PID with agent's process identifier. This variable is necessary to control the agent which also may be done from within a tox environment.

I do not know if either ssh-agent is available on Windows or it works the same way as it does on Linux/MacOS, so I do not pass the variables.

• ran the linter to address style issues (tox -e fix)
• wrote descriptive pull request text
• ensured there are test(s) validating the fix
• added news fragment in docs/changelog folder
• updated/extended the documentation