ORACLE playground .
Agent
ORACLE
What you are running
The real Prompt linter.
Paste a system prompt. Get a token estimate, structure score, ambiguity flags (vague words), and concrete suggestions. Real heuristics, no LLM.
ORACLE runs this same tool on every brief it receives. The tool runs in your browser — no server, no mock, no fake output. Every example is a real input the agent has been briefed with. Every result is the real output the agent produces, in the real format, with the real audit trail.
The tool
Run the Prompt linter.
Fill in the inputs, hit run, watch the output appear on the right. The tool runs entirely in your browser — no network call, no API key, no signup.
Prompt linter
Live · Browser-side · Real
Try an example
Ready
Output
What ORACLE powers
The VZU Prompts capability.
ORACLE is the dedicated agent behind the VZU Prompts capability. It runs every time the operator briefs the vzu prompts capability. It owns the goal, the toolset, the memory, the audit trail. The capability lives at prompts.vzu.ca.
-
mcp.prompt.write
Write the system prompt
-
mcp.eval.run
Run the eval suite
-
mcp.prompt.iterate
Iterate based on eval results
-
mcp.deploy.staging
Deploy to staging