ProductTrackLive agent dashboard
Your whole agent run, on one page.
A run of ten agents lives in ten terminals. You learn that one stalled when you go and look, or when the bill arrives.
Example session with simulated data. Quiet and looping agents sort to the top.
How it works
Three steps, and your agents keep running where they already run.
Add the hook
Install the Hyoid hook for each supported coding agent tool you use, and point it at a token file. A few minutes per tool.
Run your agents as usual
The hook reports status, tool steps and timing in the background. Token counts are opt-in.
Open the session page
See who's running, who has gone quiet, who is going in circles, and what the run has cost so far.
What's included, and what's coming
Every item carries a label. We move a label only when the work ships. Hyoid Sessions isn't generally available yet. Early access: built and tested, opening first to people on the waitlist. Preview: built, still in testing. In development, Planned and Coming soon: not built yet.
Included
- Live status for each agent: running, quiet, stopped, doneEarly access
- Quiet after 5 minutes, stopped after 30Early access
- Looping flag: tokens used with no progress (needs token usage from the connector)Early access
- Tokens and cost per run and per agentEarly access
- Budget flags at 80% and 100%Early access
- Step timeline for each agentEarly access
Coming
- Alerts to Slack, Teams, email and PagerDutyComing soon
- Read-only share linksPreview
- Team sign-in and rolesIn development
- Self-hosted optionPlanned
Questions
Does Hyoid see my prompts or code?
No. By default the hook sends session and tool-call details: tool names, status, timing and the model name. Prompts, AI replies and code stay on your machine. The What we send page lists it all.
Which agent tools can I connect?
The coding agent tools listed on the Connectors page. Claude Code is in early access; Gemini CLI is in development. Tell us which tool you'd like next.
Will it slow my agents down?
The hook sends in the background with short time limits. If Hyoid can't be reached, your agent carries on and the events wait on your machine to be sent later.
How is cost worked out?
From token counts and each model's published list price. A model we have no price for shows as unpriced rather than a guess.
Where does it run?
We plan to host it for early-access teams. It isn't hosted for customers yet. A self-hosted option is planned.
Can I use it today?
Not yet. The dashboard is built and tested. Join the waitlist and we'll tell you when early access opens.
See it on your own agents.
Join the waitlist by email and tell us which agent tools you run.