Agents
Agents Overview
Choose a guide below to get started.
- AuthSecure API keys and server-only access patterns.
- Public FillSafe patterns for publicly reachable AI-assisted form submission.
- CLICommand-line setup and reproducible local testing.
- MCPExpose tools and context over Model Context Protocol.
- SkillsDesign reusable capability modules for multi-agent workflows.
Overview
NativeForm gives AI agents a reliable way to read form data, submit structured responses, and operate with clear authentication boundaries.
New to the API? Start with Get an API Key, then read API and submit API before agent-specific patterns.
Capabilities
- GETRead submissions
/api/forms/{formId}/responsesUse the read API to pull normalized form responses into your assistant memory or pipelines.
- POSTSubmit responses
/api/forms/{formId}/responsesLet an agent convert free-form user input into structured responses through the submit endpoint.
- Manage authentication
/docs/my-api-keysIssue account-level keys, rotate compromised keys, and isolate integrations by key name.
Architecture
- Runtime boundary. Keep NativeForm keys in server runtimes only. Browser clients call your backend, not NativeForm.
- Validation. Validate incoming user input and outgoing agent tool calls before forwarding to NativeForm APIs.
- Observability. Capture request IDs, response codes, and missing field errors to improve extraction over time.
- Fallbacks. Deterministic fallbacks when the agent cannot infer required form fields with confidence.