Skip to content

SDKs and integrations

Scopebound enforces AI agent scope at the workflow level. Whatever stack you're building agents in, there's an integration that lets you validate workflows before they execute.

Available integrations

Integration Best for Status
TypeScript SDK Node.js and TypeScript codebases — direct integration with the Scopebound API Public preview
n8n Community Node n8n workflow users — drop into any workflow, validate before execution Public preview, verification pending
Python SDK Python codebases — direct integration Available, full SDK doc coming

Which one should I use?

  • Building agents in code? Use the language-native SDK (TypeScript, Python).
  • Using n8n for agent workflows? Use the n8n community node. Two commands to install.
  • Using another workflow engine (Make, Zapier, Temporal, custom)? Call the REST API directly — all SDKs are thin wrappers over the same endpoints.

How Scopebound integrations work

Every integration sends workflow definitions to a single evaluation endpoint: 

POST /v1/workflow/evaluate
The enforcement plane:

  1. Looks up the agent role you specified
  2. Translates the workflow into Scopebound's canonical shape (if it's in n8n, Make, Zapier, or another supported source format)
  3. Evaluates the canonical workflow against the role's authorization scope
  4. Returns pass/fail per compliance profile, plus a list of violations, and a cryptographically signed attestation token

The same evaluation logic runs regardless of which integration you use. The SDKs and the n8n node are convenience wrappers; behavior is identical at the API level.

What gets evaluated

By default, every evaluation includes the Production Readiness profile, which catches:

  • Out-of-scope tool calls — agents calling tools not in their allowed_tools list
  • Credential misuse — agents using credentials they aren't authorized to use
  • Approval requirements — workflows that bypass required human-in-the-loop gates
  • Idempotency violations — workflows that risk duplicate side effects

Compliance profiles can be combined: ["PRODUCTION_READINESS", "SOC1", "SOC2_TYPE_II", "HIPAA"]. Each adds its specific control set on top.

Next steps

For more on enforcement architecture and agent role configuration, see the Quickstart.