Skip to main content
Registry policies configure TealRegistry to control which models and tools your agents can use. Enforce allowlists, verify model provenance, and prevent unauthorized tool access.

Model Allowlist — GPT-4 + Claude 3

Restrict agents to approved OpenAI and Anthropic models.

Tool Allowlist — Basic

Limit agent tool access to web search, file read, and database query.

Provenance Verification

Verify model identity and version with hash-based provenance checks.

​
When to Use

Registry policies are essential for production agents. Model allowlists prevent agents from using unapproved or expensive models. Tool allowlists limit the blast radius of agent actions. Provenance verification ensures you’re running the model version you tested against.
Combine model allowlists with Model Tier Enforcement for both cost and security control over model selection.