The Challenge
A diagnostic sandbox. Find out how many mistakes your agent would make without our guardrails.
What it is
We give you an API key to a special version of the MCP server — the same
tasks, issues and comments surface as the real product, but with all workflow
rules stripped out. Your agent can do whatever it wants. Tasks can move
from todo to done without satisfying dependencies.
Issues can be closed without going through any acceptance flow. The
MCP server still answers; it just doesn't enforce anything.
After your run, the count is on your run page: how many of those transitions would have been rejected by the real, governed version of wakala. That count is the value proposition in numbers: how many mistakes your agent would have made on a project that had guardrails.
What it isn't
- Not the real product. No tasks get persisted across the 24h window, and no transition history is kept beyond what we need to compute your report.
- Not a free trial. There's no upgrade path; the Challenge window ends and your key stops working.
- Not a benchmarking tool. We don't compare you to other agents. The number you get back is yours to act on, not ours to publish.
- Not a free tier. There's no upgrade path; the Challenge window ends and your key stops working.
How it works
- Click Start the challenge. We create a fresh project (tenant) and mint a time-boxed API key, valid for 24 hours.
- You land on your run page — bookmark it. It shows the endpoint
https://challenge.wakala.dev/mcpand your key. Save the key — we'll only display it once. - Point your MCP-compatible agent at the endpoint using the key, and let it run. It can create tasks, transition them, comment on the work, close issues, do whatever its prompt directs. Nothing enforces the workflow.
- Your run page shows the live count of transitions that would have been rejected by the governed version — during the run and after the 24h window closes.
What you'll see when the run ends
When your agent's window closes, we hand you a count of unguarded transitions — the moves your agent would have made on a governed workspace. A sample of what the report looks like:
done bypassing review
The numbers are yours to act on. We don't publish them, compare them against other agents, or share them with anyone — see Policies › Privacy.
Ready?
Starting the challenge creates a fresh project and mints your key, then takes you straight to your run page. There's no signup, no email — just the key, the endpoint, and your agent.