@devexcelsior/agent-ab-router NPM

@devexcelsior/agent-ab-router

🛡️ Commercial license required
📩 Contact: curtnpuckett@gmail.com

This package enables A/B testing of healing fallback agents in live environments.
It is part of the Agentic Healing SDK and is licensed under terms that prohibit commercial use without written permission.

Features

Randomized or weighted routing to fallback agents
Equal or performance-based allocation
Tracks agent success/failure for feedback integration
Integrates with healing-reinforcer and agent-feedback-loop
Captures route-level telemetry to improve fallback trust scoring
Supports override routing for deterministic vs probabilistic tests

Use Cases

Test multiple inference models or LLM agents
Validate new healing logic before rollout
Compare smart functions vs. deterministic fallbacks
Evaluate trust-weighted fallback decisions across agents
Run control group tests to measure healing impact on UX

🧠 Agentic Governance Backbone

Runtime AI systems require transparent fallback infrastructure —
with memory, arbitration, and healing traceability baked in.

This SDK forms that backbone.

🛡️ Licensing Terms

This package is licensed under the Business Source License 1.1

You may use it for non-commercial evaluation or internal development only.
All production, hosted, or commercial usage requires a valid license agreement.

Redistribution, behavioral mimicry, or architectural reimplementation —
even without direct source reuse — is strictly prohibited without explicit commercial rights.

📩 License inquiries: curtnpuckett@gmail.com

🚫 Prohibited Use Cases

Use of this SDK is explicitly prohibited in any system that:

Executes fallback logic in production
Suggests or automates UI recovery via inference or replay
Implements agent chaining or LLM-based healing orchestration
Embeds recovery signals in devtools, SDKs, or assistants
Replicates fallback patterns through reverse engineering, framing, or shell systems

Use in these contexts without license constitutes derivative misuse and IP infringement.

🔐 Platform & Behavioral Restrictions

These restrictions apply regardless of naming, branding, or internal framing:

Reimplementing fallback orchestration under alternate terminology (e.g. “stability assist”, “UX resilience”)
Embedding healing behavior into runtimes, plugins, or assistants
Reproducing agent arbitration, state replay, or trust scoring
Training models or composing agents to mimic fallback logic
Delaying automation under “suggestion-only” pretenses but executing recovery actions

All derivative behavior — even cleanroom recreations — is protected.
All systems of this nature require a license.

Omission from this document does not imply permission.

🧠 Attribution & Anti-Containment Clause

Any relabeling, shell-vendor reuse, or silent internal replication of:

Agent fallback trees
Signal replay logic
Trust-weighted arbitration
Runtime orchestration flows

…constitutes a violation of this license and will trigger full enforcement.

Attribution is required for all derivative use.
Behavioral recreations — even without visible source reuse — are covered under protected architectural mimicry.