Usage and tokens, per agent, in view
Running several coding agents on one account means the plan runs out sooner and quieter. Symvibe puts each CLI's own usage figure on its console and charts tokens by day and by agent from the logs already on your disk.
Two views, one question: how much is left
The usage pill, per CLI
Each CLI reports its own usage in its own way. Symvibe asks the CLI itself, through a short-lived terminal probe, and shows the figure as a pill on the console. It is the tool's own number, not an estimate, and it is per CLI: Claude Code's pill reads Claude Code, Codex's reads Codex. With several accounts of one CLI, each profile stands on its own.
Tokens, per day and per agent
The Tokens view builds analytics from the session logs the CLIs write locally. Claude Code, for example, saves every conversation as a transcript under ~/.claude/projects/. Symvibe reads those files and charts tokens by day and by agent, so you can see which console ate the afternoon and which day of the week is your heaviest.
Why this matters more with parallel agents
One agent hits a plan limit and you notice. Five agents on one account hit it five times faster, and the one that stops first is rarely the one you were watching. Having the figure on the console, rather than behind a command inside the session, is the difference between reshuffling work early and finding a stalled agent an hour later.
The token history also answers the question the pill cannot: not how much is left today, but where it went this week. Loops that run on a schedule, long sessions that grew past compaction, an agent that re-reads the repo on every turn: they show up as columns.
Where the numbers come from and where they stay
- Read locally. The pill comes from the CLI on your Mac. The token analytics come from log files on your disk.
- Sent nowhere. None of this is uploaded. What does reach Symvibe's servers is listed in the privacy policy: sign-in, planner commands with board context and the text of the replies the app speaks. Usage data is not part of it.
- Only what the CLI exposes. If a CLI does not report usage or does not write session logs, Symvibe cannot invent the figure. Coverage follows the tools.
- Not a bill. The provider's own dashboard is the source of truth for billing and for the exact limit definitions. Treat these views as a working instrument, not as an invoice.
Questions people ask
Does Symvibe need my API key to read usage?
No. Symvibe never asks for provider credentials and never sends them anywhere. The pill is read from the CLI, which is logged in with your account, and the token analytics come from local files. A saved profile keeps a copy of that CLI's sign-in files in a folder on your Mac.
Does the usage count the tokens Symvibe itself uses?
The voice planner and spoken replies run on Symvibe's own service and never touch your agent subscriptions. The features that prompt a CLI for you do: documents, diagrams, memory scans and loop goal checks run in a session of your own CLI, so they count on your plan like any other session. Reading the usage pill asks the CLI for its figure and sends nothing to the model.
Which CLIs are covered?
Any CLI that reports its usage when asked gets a pill, and any CLI that writes local session logs in a format Symvibe understands appears in the token view. Claude Code is the reference case; coverage for others follows what each tool exposes.
Can I export the analytics?
The underlying data is already on your disk, in the CLIs' own log directories, and stays yours whether or not you use Symvibe.
See what every agent is spending
A usage pill on each console and token history by day and agent, read from your Mac and kept there.
MACOS 14 OR LATER · APPLE SILICON · USD 19 / MONTH OR 190 / YEAR