Compatibility
Model-agnostic by design
Clarify provides an execution protocol—not a model dependency. Any agent that can call tools can integrate.
Model-agnostic architecture
Clarify exposes a standard execution API (Quote / Execute / Receipt). We don't depend on any specific LLM provider or agent framework. If your agent runtime can make HTTPS requests or call MCP servers, it can use Clarify.
- No vendor lock-in—switch models without changing your commerce integration
- Consistent receipts and consent artifacts regardless of which model initiates the transaction
- Same trust layer works for cloud-hosted or self-hosted models
Integration methods
HTTP API
Standard REST endpoints. Any runtime that can make HTTPS requests can integrate.
MCP Server
Model Context Protocol support for agents using MCP-compatible tool calling.
Function calling
Wrap Clarify endpoints as tool definitions for LLMs with native function calling.
Compatibility matrix
Example agent runtimes and their integration status. This is not an exhaustive list.
We list provider names as examples to illustrate the types of stacks that are compatible by design. 'Compatible by design' means Clarify's architecture supports integration through standard protocols (HTTP, MCP, function calling)—not that we have a commercial partnership with these providers.
| Provider / Stack | Integration path | Status |
|---|---|---|
| OpenAI-compatible agents | HTTP API / Function calling | Compatible by design |
| Google LLM stack (e.g., Gemini-based agents) | HTTP API / Function calling | Compatible by design |
| Amazon Bedrock-based agents | HTTP API | Compatible by design |
| Anthropic Claude agents | HTTP API / MCP / Tool use | Compatible by design |
| Self-hosted / open-weight models | HTTP API | Compatible by design |
| LangChain / LlamaIndex agents | HTTP API / Custom tools | Compatible by design |
What 'Tested' means
A reproducible demo or integration test exists in our repository. 'Compatible by design' means the architecture supports integration, but we haven't published a specific test harness yet.
Vendor names are provided as examples only. Clarify is not affiliated with, endorsed by, or officially partnered with the providers listed above. All trademarks belong to their respective owners.
// Any agent that can call tools can integrate
const quote = await clarify.quote({
cart: items,
buyer: { email: "[email protected]" }
});
// Model-agnostic: HTTP, MCP, or function calling
const result = await clarify.execute({
quoteId: quote.id,
consent: userConsent
});Ready to integrate?
Request pilot access and we'll help you connect your agent runtime.