Developers

Integrate governance as a boundary, not a prompt convention.

TenantSage is designed to sit between verified request context and AI retrieval or execution. Integrations consume governed artifacts rather than reconstructing authorization independently inside model or vector-search code.

Core integration concepts

AuthoritySnapshot
GovernanceContext
EEB reference
RetrievalReceipt
ValidationApproval
ExecutionIntent
ExecutionReceipt
Durable intent and completion proof
Replay lineage

API overview

A conceptual public surface, not proprietary implementation disclosure.

Endpoint names describe integration responsibilities and may evolve before a stable public API release. Internal policy logic, database mechanisms and privileged topology are intentionally not exposed here.

POST/authority/resolveResolve current authority from verified request context.
POST/governance/decisionEvaluate governance state and produce a governed decision.
POST/eeb/materializeConstruct and seal the eligible evidence boundary.
POST/retrieval/searchRetrieve only within the sealed EEB.
POST/execution/intentPersist an execution intent bound to ValidationApproval.
POST/execution/completeRecord verified completion evidence.
GET/execution/replayRetrieve replay evidence for a governed execution.

Developer guides

Multi-tenant integration
Explicit scope binding
RLS enforcement
Cache invalidation against authority/policy state
Durable receipt lifecycle
Replay verification
Negative-path testing

Synthetic examples

Demonstrate behavior without leaking customer or policy internals.

Developer examples should use synthetic tenants, scopes and evidence. Reference flows include governed retrieval, approval-bound execution and negative tests showing explicit denial when scope, authority, policy freshness, EEB integrity or durable receipt requirements fail.

SDK roadmap

TypeScript — plannedPython — plannedGo — planned

No SDK availability is implied until a versioned release and corresponding conformance evidence are published.