Back to Shree AI OS
SHREE AI OS · SDK
01

Build through
the SDK.

The Shree SDK provides the developer-facing layer for connecting applications with the platform runtime and reusable intelligence capabilities.

v1.0.6 · 10 VERIFIED SDKS

One developer
interface.

Applications should not need to understand the internal implementation of every intelligence capability they use.

The SDK acts as the developer-facing boundary between an application and the underlying Shree AI OS platform.

APPLICATION → PLATFORMSDK
YOUR PRODUCTApplication
DEVELOPER LAYERShree SDK
PLATFORMShree Runtime

Connect to
platform intelligence.

The SDK is designed as the developer-facing entry point for platform capabilities. Applications can consume reusable intelligence without coupling themselves to internal systems.

01
MEMORY

MemorySDK (client.memory())

Persistent episodic context, semantic recall, and thread-safe session caching scoped strictly per tenant.

02
KNOWLEDGE

KnowledgeSDK (client.knowledge())

Hybrid RRF vector retrieval (PostgreSQL pgvector HNSW + GIN FTS) and K0.6 Autonomous Knowledge Acquisition.

03
PLANNING

PlanningSDK (client.planning())

Deterministic goal decomposition into topological DAGs with typed constraints and validation criteria.

04
REASONING

ReasoningSDK (client.reasoning())

Structured thought evaluation, premise verification, and conflict-free evidence graph construction.

05
REFLECTION

ReflectionSDK (client.reflection())

Post-execution reflection, outcome analysis, adaptive threshold tuning, and tenant-scoped performance analytics.

06
INFERENCE

InferenceSDK (client.inference())

Deterministic hypothesis scoring, tradeoff evaluation, and calibrated confidence tier assessment.

07
IDENTITY

IdentitySDK (client.identity())

Actor resolution, tenant isolation enforcement, workspace identification, and profile management.

08
EXECUTION

ExecutionSDK (client.execution())

Action plan execution protected by the fail-closed authorization gate (graphPermissionManager).

09
PROJECT

ProjectSDK (client.project())

JavaParser Java 21 AST analysis, structural impact analysis, in-memory patch generation, and workspace scanning.

10
SETTINGS

SettingsSDK (client.settings())

Zero-downtime hot-reload BYOK credential storage, provider routing thresholds, and runtime configuration.

From application
to intelligence.

The developer workflow remains intentionally layered. Each component has a clear responsibility while the platform handles the underlying intelligence infrastructure.

01

Application

Your product and domain-specific experience.

02

SDK

Developer-facing interface to Shree AI OS.

03

Runtime

Platform lifecycle, context, and execution.

04

Intelligence

Reusable intelligence capabilities and kernels.

Simple for
developers.

The SDK follows the same architectural principles as the platform: clear boundaries, reusable capabilities, and interfaces designed for long-term evolution.

01

Stable Interfaces

Applications should communicate through defined interfaces instead of depending directly on internal implementations.

02

Reusable Capabilities

Common intelligence infrastructure belongs to the platform and should be reusable across applications.

03

Clear Boundaries

The SDK separates application concerns from runtime and intelligence infrastructure.

CURRENT PLATFORM STATE

The interface evolves
with the platform.

VERSIONV1
STATUSFOUNDATION
FOCUSPLATFORM INTERFACES
SHREE AI OS · SDK

Build the application.
Reuse the intelligence.

Explore the documentation and architecture behind the Shree developer platform.