Developer Preview · NEES Core V3 · Nainacore Emotional Tech
AI agents can send emails, update databases, access files and call external APIs. But just because an agent can perform an action, should it be allowed to?
NEES is an independent AI action governance platform. It helps applications evaluate an agent's authority, policies, permitted scope and current governance state before consequential actions proceed.
NEES is not an LLM, chatbot or replacement for your agent framework. It is a governance boundary that helps make supported AI actions controlled, traceable and accountable.
Imagine an agent asked to send a confidential report. It knows how to send email—but:
- Is it authorized to send messages?
- Is that recipient within the permitted scope?
- Does policy require approval or block this operation?
- Can we inspect the decision and what actually happened?
NEES provides governance checks and evidence. An ALLOW decision is not proof of execution: the owning application must still validate its actual target, perform the effect safely and report the real outcome.
User / Application -> AI Agent -> Proposed Action
|
v
NEES Governance
Authority · Policy · Scope
Tool permissions · State
|
Governance result
/ \
Not authorized Authorized
| |
Do not execute App rechecks context
|
Executes & verifies
|
Reports outcome
|
Evidence & receipt
NEES governs supported operations; your application owns execution and verification.
| Method | Typical use |
|---|---|
| Python SDK | Python agents and backends |
| REST API | Applications in any supported language |
| Gateway / Proxy | Mediated HTTP service and tool calls |
| MCP Adapter | Supported MCP-based agent/tool integrations |
| Framework Adapters | Supported agent framework hooks |
Availability and configuration depend on the specific connector and deployment.
- Sign in at app.nees.cloud and select/create an organization.
- Set up your application: Project → Runtime → Environment. Issue a runtime credential and keep it server-side.
- Register your real action with its resource, operation, capability and side-effect metadata.
- Publish governance for that action through an authorized workflow.
- Open Infrastructure → Assessment configuration manager. Select Registered action → Published scope → Integration method and click Resolve existing assessment.
- Integrate and test: submit → qualify → start → report the actual outcome; inspect operation, evidence and receipt.
The Dashboard exports the exact version-bound assessment commitments for the selected existing governance baseline. It does not export your API key or grant permission by itself.
Important: Publish baseline & resolve changes governance records and may overwrite custom policy commitments. Use Resolve existing assessment for read-only configuration lookup.
| Guide | What you'll find |
|---|---|
| Quickstart | First integration walkthrough |
| Dashboard assessment configuration | Actions, published scopes, exports and safety |
| Complete integration guide | Supported lifecycle and boundaries |
| Integration patterns | SDK, REST, Gateway, MCP and adapters |
| Python SDK usage | V3 Python client and CLI |
| Access | Account and credential guidance |
| Troubleshooting | Common errors and recovery |
| Production checklist | Production acceptance gates |
| Security and scope | Secrets, limits and responsibilities |
| Case studies | Governance Lab and NainaSOS learnings |
Python SDK note: Current V3 governance SDK uses the nees-sdk 3.x release-candidate family. The older nees-core-sdk 0.1.x chat interface is not a compatible V3 governance client. See SDK usage.
We welcome AI agent developers, backend engineers, researchers and testers to explore governance for their supported actions, test allow/deny paths, inspect evidence and share integration feedback.
Preview scope: Evaluation and integration testing. Independent developer onboarding and deployment-specific production readiness still require validation. Never publish real API keys, organization credentials, customer data or sensitive governance traces.
© 2026 Nainacore Emotional Tech.