@intended/sdk
Core SDKTypeScript / Node.js
Full-featured client for the Intended platform API. Intent submission, execution pipeline, escalation handling, audit queries, and policy management.
- Full execution pipeline
- Typed API responses
- Automatic retry with backoff
- Escalation polling
- Audit chain queries
@intended/verify
Core SDKTypeScript / Node.js
Lightweight Authority Token verification library. RS256 signature validation, claim checking, nonce verification. Designed for any service that consumes a token and needs to independently confirm it was authorized.
- RS256 signature validation
- Claim integrity checks
- Nonce consumption tracking
- Expiry verification
- Required claim matching
@intended/enforce
Core SDKTypeScript / Node.js
Enforcement SDK for decision request and enforcement flows. Request authority, handle escalations, and execute with tokens in a single workflow.
- Decision request flow
- Escalation wait/poll
- Token-gated execution
- Retry with authority
- Webhook integration
@intended/connector-sdk
IntegrationTypeScript / Node.js
Build custom execution adapters for any target system. BaseAdapter handles token verification, nonce consumption, and audit recording so you implement just execute() and testConnection().
- BaseAdapter scaffolding
- Automatic token verification
- Nonce consumption handling
- Audit recording hooks
- Test harness for adapters
@intended/cli
ToolCLI
Command-line interface for intent simulation, token inspection, policy deployment, audit chain verification, and evidence export. Useful for CI/CD pipelines and operator workflows.
- Intent simulation
- Token inspection
- Policy deployment
- Audit chain verification
- Evidence export
@intended/mcp-gateway
RuntimeTypeScript / Node.js
Authority gateway for Model Context Protocol servers. Intercepts every MCP tool call, classifies it against the Open Intent Layer, evaluates policy, and only forwards authorized invocations.
- Drop-in MCP proxy
- Per-call authority evaluation
- Fail-closed enforcement
- Open Intent Layer classification
- Audit trail per call
@intended/k8s-admission
RuntimeKubernetes
Kubernetes admission controller that applies the Intended authority boundary to cluster mutations. Every apply, delete, and scale is classified, evaluated, and authorized or blocked.
- ValidatingAdmissionWebhook
- Helm chart included
- Per-namespace policies
- Fail-closed on API unreachable
- Audit trail per mutation
@intended/openshell-adapter
RuntimeTypeScript / Node.js
NVIDIA OpenShell / NemoClaw integration. Compile Open Intent Layer classifications and Domain LIM signals into policy YAML for the NVIDIA agent runtime stack.
- OpenShell policy compiler
- Domain LIM integration
- NemoClaw-ready output
- Per-environment policy packs
- CLI-accessible via @intended/cli
intended-python
Language SDKPython
Python client for the Intended platform. Intent submission, Authority Token verification, audit queries, plus ready-made guards for LangChain, Pydantic AI, OpenAI Agents SDK, and MCP.
- Intent submission
- Token verification
- LangChain integration
- Pydantic AI integration
- OpenAI Agents SDK integration
- MCP middleware
intended-go
Language SDKGo
Go client for the Intended platform. Designed for infrastructure tooling, Kubernetes operators, and Go-based automation systems. Currently in preview — see ADR-006.
- Intent submission
- Token verification
- Context-aware timeouts
- gRPC-ready types
terraform-provider-intended
Infrastructure as CodeTerraform
Terraform provider for managing Intended platform tenants, policies, connectors, and authority boundaries as code. Roadmap (ADR-008) — types scaffolded; resource implementations land with the first IaC-driven design partner.
- Tenant management
- Policy-as-code
- Connector provisioning
- Authority boundary config
- Plan/apply integration