MCP-first. Never MCP-only.
Your stack decides how RegAlytics® shows up. SWORD fronts four mechanisms with one consistent contract: an MCP server for AI-native workflows, a REST API for database-level access, webhooks for event-driven stacks, and flat files for batch pipelines. Same production feed through each. Same freshness. No second-class copy.
The four delivery mechanisms
MCP server. One install puts the full regulatory corpus inside any MCP-compatible AI tool — whether that's a production agent or your team's Claude Desktop. No custom wrappers, no glue code.
REST API. Paginated search, rich query options, and full taxonomy lookups across every endpoint. The workhorse for product-side integrations.
Webhooks. Event-driven delivery for stacks that react to regulatory change in real time. Subscribe once and stay in sync.
Flat file. For backfills and one-off pulls — when you need a specific slice of data without standing up an integration. Scheduled exports to S3 on whatever cadence makes sense.
Which mechanism for which use case
Most teams settle on one primary mechanism and keep a second one in reserve. The matrix below is the shortest honest answer we can give to “which should we use?”
| Use case | Recommendation | Why |
|---|---|---|
| AI agent, LLM workflow, or MCP-compatible chat client (Claude, ChatGPT, Cursor, etc.) | MCP server | One install puts the full regulatory corpus inside your AI tool as a native capability. Works equally well for production agents and individual analysts using AI clients. |
| Product surface with search, faceting, or drill-down | REST API | Database-level access with paginated, filterable endpoints for every taxonomy and alert query. |
| Event-driven stack that reacts to regulatory change | Webhooks | Stay in sync without polling. Push events delivered as new alerts are published. |
| Data warehouse or BI tool needs continuous regulatory data | REST API + flat file backfill | REST handles ongoing sync into the warehouse; a one-time flat-file pull seeds the historical record. |
| Need a specific slice of data once, without an integration | Flat file | Scheduled export of exactly the records you need, dropped to S3. No API setup. |
| Hybrid product: live integration plus historical analytics | MCP or REST, with flat file as needed | Pick the live mechanism that fits your stack. Pull flat-file extracts for one-off analyses. |
See the docs from the inside.
The fastest way to learn SWORD is to use it. A trial account gets you the full docs, working credentials, and test tokens to put any of the four mechanisms through real integration paces – before procurement, before a single contract conversation.