Coding with Entevra ADE

Choose Sign in in Entevra ADE. Your browser opens entevra.com and asks you to allow the ADE; sign in there if needed, check the account and choose Allow, and the ADE finishes by itself. If the browser cannot return to the ADE, it shows a code to enter at entevra.com/device instead.

Settings → Account → Entevra ADE lists the computers signed in to your account and can sign them all out.

Tell the agent what you want to change, which files matter and any constraints it should follow. The ADE uses the same account, history, models and budget as the web chat.

Work in small steps. Review proposed edits and Git diffs, and keep checkpoints so you can return to a known state.

• Include the expected behavior in your request.
• Check changes outside the files you expected to touch.
• Use your project’s tests to verify the result.

The agent can run build and test commands, which can affect your files and environment. Read each command and understand its scope before allowing it, and remember that some modes do not stop to ask. Use a separate copy of unfamiliar code.

The code, file contents and command output the agent works with are sent to the model you chose. Keep credentials and unrelated private files out of its reach. Agent runs never send your uploaded files or memory notes.