Sign in and opt in
Kenaz, a Kameas AI product.
Signing in connects this device to your organization's Kameas Fleet. Signing in registers this device with your organization's Fleet but doesn't share any usage: you choose that in a separate step, after you've read the notice.
Sign in
kenaz login
kenaz login opens your browser to sign in with your Kameas account.
Sign-in returns to a one-time listener on 127.0.0.1 (a random port) in the kenaz process.
If no browser can be opened, or with kenaz login --no-browser, kenaz prints the sign-in link instead and keeps waiting.
--env prod|stage|dev picks the Kameas Fleet environment (default prod).
In Claude Code, /kenaz:login asks Claude Code to run kenaz login for you.
Signing in also starts the local relay. kenaz init runs kenaz login for you if you aren't signed in yet.
If your organization isn't in the early-access program, kenaz login stops with:
kenaz: Kenaz for Claude Code and Codex is in early access. Contact sales@kameas.ai to join.
Choose what to share
Right after your first sign-in, kenaz login shows the Agent Telemetry Notice and asks what to share.
You can come back to it at any time with:
kenaz optin
The notice lists the classes your organization's plan includes as numbered checkboxes, all unchecked; classes your plan doesn't include are shown greyed out.
- Share selected: type the numbers of the classes to share (for example
1,2), then confirm withy. - Not now: press Enter. Nothing is shared, and you can choose later with
kenaz optin.
Opting in only works in your own terminal.
kenaz optin refuses answers from a pipe or a script, so an agent can't opt you in, and there's no command-line flag that opts you in.
To turn classes off without the notice:
kenaz optin --class harness.tool_calls --off # one class
kenaz optin --all-off # every class
The class names are harness.usage_counts, harness.tool_calls, harness.errors and model_call_telemetry.
kenaz status shows which classes are on.
What each class sends
This is the Agent Telemetry Notice that kenaz login and kenaz optin show, word for word.
{Organization name}, {harnesses}, {device display name} and {harness versions} are filled in with your details when it's shown.
The full public description is at kenaz.kameas.ai/telemetry.
Share agent usage with {Organization name}?
Kenaz can send {Organization name}'s Fleet a summary of how you use {harnesses} on this device. Nothing is shared until you choose below, and you can change your choice at any time with
kenaz optin.What can be sent, by class (classes your plan doesn't include are shown greyed out):
- Usage counts: when a session starts and ends, how long it lasts, how many tokens it used and its estimated cost.
- Tool calls: which kind of tool was used (for example Bash, Edit or Read, or simply "MCP" or "other" for tools you or your organization added), whether it worked and how long it took.
- Errors: the kind of error (for example "rate limited" or "network") and whether it was recoverable. Never the error message.
- Model calls: for each request to an AI model, the model's name, the provider, token counts, estimated cost, timing, why the response ended, and a coded session reference that groups calls from one session without revealing the session itself.
Attached to every record: your Kameas account, your organization, this device's ID and display name ("{device display name}"), and which tool sent it ({harness versions}).
Never sent: your prompts, AI responses, tool inputs or outputs, file contents, source code, file paths, folder or repository names, error messages, or the names of tools and servers you added. This filtering happens on your device, in code you can read at github.com/kameas-ai/kenaz-plugin, and Fleet rejects anything outside this list again on arrival.
Who sees it: {Organization name}'s Fleet administrators see team-level totals, for example sessions, cost and tool use per day. In a small team, totals may effectively show your own activity. They can see that this device is connected and when it last checked in. Your organization has agreed not to use this data to evaluate individual members. Kameas processes it on your organization's behalf.
How long it is kept: for as long as {Organization name} uses Kameas Fleet, unless you opt out or leave. Opting out of a class deletes your stored records of that class within 72 hours.
Your rights: your organization is responsible for this data. Contact your organization's administrator to access, export or delete it. Kameas's Privacy Policy, section 13, explains more.
Separately, Kenaz may sync skills and team context you choose to install, publish or add. What you publish is shared with your organization as you would expect. Kenaz never changes or blocks what Claude Code or Codex does.
Depending on your organization, the notice you're shown may have a different Who sees it paragraph.
The text kenaz login and kenaz optin show you is the one that applies.
Your device's name
Records carry this device's display name, which you control:
kenaz device # show it
kenaz device rename "Work laptop"
kenaz device rename --use-computer-name
Where your sign-in is kept
Sign-in tokens, and the key that encrypts uploads waiting to be sent, are kept in the OS keychain: the macOS Keychain, the Windows Credential Manager, or the Secret Service on Linux.
On a machine with no usable OS keychain, kenaz keeps them as files in a credentials directory under its state directory, readable only by you (directory mode 0700, files 0600).
In that case they're protected only by file permissions.
kenaz status and kenaz doctor show which store is in use.
Sign out
kenaz logout
kenaz logout signs out and deletes data on this device that was waiting to be sent to Kameas Fleet.
It keeps your memory on this device, and stops memory sync.
Claude and Claude Code are trademarks of Anthropic, PBC. OpenAI and Codex are trademarks of OpenAI. Kenaz and Kameas are trademarks of Kameas AI, Inc. Kenaz for Claude Code and Codex is an independent product of Kameas AI, Inc. and is not affiliated with, endorsed or sponsored by Anthropic or OpenAI.