Trace the activity
Inspect individual sessions and review cost estimates by project and day, drawn from your local transcripts.
Agent operations, under your control
Let your agents run.
Keep them
in check.
A control dashboard for your Claude Code agents. See what they cost, block tool calls over your limits, and find automations that quietly stopped working.
Runs on your machine. Full source included. One-time purchase.
YOUR AGENT OPERATIONSILLUSTRATED EXAMPLE
Coding sessionOver its configured budget
LIMIT REACHED
Scheduled reportExpected output is missing
STALE
Project activitySessions and estimated spend
VISIBLE
THE FORGE / AGENT OPSCommand Center
01 / Block02 / Investigate03 / Resolve
Your machine. Your data. Your call.
01 / When you need it
More autonomy.
Fewer blind spots.
You have agents doing real work. This is where you go to understand what’s happening—and decide what happens next.
COST CIRCUIT BREAKERExample workflow
Give the next tool call
a stopping point.
Set a budget for a session, project, or day. With the Claude Code hook connected, the breaker checks recorded spend before a tool call and blocks it when the limit is breached.
- 01 Set your budget
- 02 Hook checks recorded spend
- 03 Over the limit? Block the tool call
A tool-call guard, not a guaranteed billing cap. It cannot stop model generation or API calls outside Claude Code.
02 / Inside the product
Start with what
needs attention.
A guided workspace for reviewing agent work: see what needs attention, open the evidence, and choose a next step. Try it with sample data in the live demo.
Open the interactive demo
Intervene deliberately
Block tool calls with a budget rule. Kill or restart a process only when it’s on your allowlist and you confirm.
Keep a record of the fix
Track errors through open, monitoring, and resolved states, with notes that keep the investigation in one place.
03 / How it works
Lives alongside
your agents.
Install it where your Claude Code sessions run. You own the application, the configuration, and the data it reads.
Check if it fits your setup-
01
Download and install
Get the source ZIP and license after checkout. Run the installer on macOS or Linux with Python 3.11+.
-
02
Connect what you run
The dashboard reads local Claude Code transcripts. Register your scheduled jobs and connect the tool-call hook to enable budget enforcement.
-
03
Set your rules. Open your dashboard.
Choose session, project, and daily limits. Review automation health and decide which processes can be stopped or restarted.
04 / Make it yours
Your agents.
Your control room.
For developers and small teams running multiple Claude Code sessions or scheduled agent work.
Want to talk it through first?
Tell us what you run and what you need to keep an eye on. We’ll help you work out whether it fits.
Talk to Pawel pawel@jock.pl
AGENT OPS COMMAND CENTERTEAM LICENSE
$499 USD / one-time
Self-hosted software. Full source. Yours to run.
- Session and project cost dashboard
- Session, project, and daily budget breakers
- Automation health, lock, and heartbeat checks
- Error registry and allowlisted process controls
- Source code, installer, and setup documentation
- Offline-verifiable license, no recurring fee
Secure checkout with Stripe. Download on the confirmation page.
Checkout couldn’t open. Please try again or
email Pawel for help.
Already own the Agent Efficiency Kit?
Email your receipt
for a $49.99 upgrade credit before purchasing.
Before you decide
A few useful
details.
Try the demo. Know the limits.
Ask us anything that’s missing.
Is this a hosted service or something I install?
It’s software you install on your own machine. The local dashboard reads your Claude Code transcripts and the automation registries you configure. Your session data stays with you. The hosted demo uses synthetic data.
Will this work with my agent setup?
The supported setup is Claude Code on macOS or Linux, with Python 3.11 or newer. It reads transcripts available on that machine; it does not automatically connect agents on other computers. There’s no native Windows installer. If you use another agent framework or need multiple hosts, tell us about your setup before buying.
Does the breaker guarantee a spending cap?
No. With the PreToolUse hook installed, it can block a Claude Code tool call when recorded spend has breached your configured limit. It does not intercept model generation or API requests outside Claude Code. Cost is estimated from transcripts and a model pricing table, not your invoice. Unknown models and malformed transcripts can leave gaps; mixed-model cost attribution is approximate.
Does it monitor continuously or send alerts?
Automation health checks run when the dashboard or its API is requested. There is no separate background watchdog or built-in push alert for the dashboard going down. For that, add an external uptime check. Jobs must be registered with the expected signals before Command Center can assess their health.
What happens after I pay?
The confirmation page provides your license and download. The ZIP includes the source, installer, and buyer documentation. Save your license and download before closing the page; email delivery is a backup. Your purchased version has no subscription expiry. Fixes and improvements to this release are provided through re-download with your license.
What if it doesn’t do what this page says?
Try the live demo first. If the installed product doesn’t do what’s described here, email Pawel within 14 days for a refund request.
How does it relate to the Agent Efficiency Kit?
The $49.99 Agent Efficiency Kit focuses on measurement hooks and a local dashboard. Command Center adds budget enforcement, automation health checks, and process controls. Kit owners can request a $49.99 credit toward Command Center before checkout.
See your control room
before you make it yours.
Explore the live demo
No signup. Synthetic data. The real dashboard.