Hyper Agent Integration Kit

0,00 €

Also purchasable with 250 Credits. Log in to use Credits.

Categories: ,

Description

Please note this is a Beta product. This product is as powerfull as the AI model you choose. I advise Claude Opus. It requires the use of Github Copilot or Claude.
This product can only be used with all features by subscribers on all tiers for the moment on Games by Hyper.
Lol on the AI image, haha. Apologies. Did not had time to make a custom one. Focused on the development side at the moment.
Since it is Beta, I don’t yet know how I want to ship this. Currently I want to give access to our subscribers via the credits system.  I will likely change the method. E.g. subscribers only, maybe on Fab. Maybe a freemium model. I have not yet decided. As long as the community benefits from it and I get some revenue to improve the system.
Have fun! Greetings, Eric

Hyper Agent Integration Kit

v0.6.0
Unreal Engine 5.7 • Windows 64-bit • Prebuilt • No compilation

Visual Overview

You (Editor Chat)
        │
        │  "Create a health pickup"
        ▼
AI (Copilot / GPT / Claude)
- Interprets intent
- Plans execution steps
        │
        ▼
MCP Server (local .exe)
- Translates AI → Unreal commands
        │
        ▼
UE Plugin (Editor)
- Executes 300+ tools
- Builds assets live
  • Local TCP (no cloud)
  • Full roundtrip loop with feedback
  • Continuous execution chain

Core Proposition

  • Direct AI control over Unreal Editor
  • Execution, not suggestion
  • Real-time asset creation

Result:

  • Intent → system implementation
  • No manual graph wiring
  • No context switching

 

Differentiation

Not:

  • Code generator
  • Autocomplete
  • Assistant overlay

Is:

  • Editor execution layer
  • AI-driven tooling system
  • Full pipeline automation

Capability Surface

Coverage

  • 300+ tools
  • 27 categories
  • 35+ one-click workflows

Domains

  • Blueprints (full graph control)
  • Actors & level design
  • Materials & rendering
  • UI (UMG)
  • AI systems (BT, EQS, GAS)
  • Animation, Control Rig, IK
  • PCG
  • Niagara
  • Input
  • Networking
  • Asset management

Workflows

Two entry modes:

1. Prompt-driven

  • Natural language
  • Multi-step execution
  • Adaptive (error handling)

Example:

  • “Create sprint system with stamina drain”

2. One-click workflows

  • Predefined templates
  • Instant generation
  • Stored in /Game/_HyperAgent/Examples/

UI System (In-Editor)

7 tabs:

  • Chat
    • Execution stream
    • Tool calls visible
  • Learning
    • Project scan
    • Builds knowledge base
  • Rules
    • Constraint injection
    • Deterministic behavior shaping
  • Dictionary
    • Structured asset index
    • Direct navigation
  • Workflows
    • Prebuilt generators
    • Category-based
  • Info
    • Internal documentation
  • License
    • Activation + validation

Context Model

  • Full project scan
  • Injected into AI context

Includes:

  • Blueprints (functions, vars, components)
  • Structs / enums
  • Widgets
  • Gameplay tags
  • Data tables
  • Dependency graph

Effect:

  • AI operates on real project state
  • Eliminates hallucination risk (partially)

Permission System

Per category control:

  • Always allow
  • Ask
  • Deny

Categories:

  • Create / Edit / Delete
  • Compile
  • Spawn
  • Python
  • Console

Default:

  • Python disabled

Safety Model

Proven

  • Transaction system (Undo-safe)
  • Blueprint compile validation
  • Pin compatibility checks
  • Auto-save before batches
  • Full audit logging

Strongly Likely

  • Crash containment via SEH
  • Retry logic for dependent ops

Effect

  • Failure → recoverable
  • Editor stability preserved

Performance Model

  • 100% local execution
  • No network latency
  • No external APIs

Implication:

  • Deterministic speed
  • Privacy preserved

Installation

  • Copy plugin → /Plugins/
  • Open project
  • Tools → Hyper Agent Chat
  • Start session

License required via:

Strategic Impact

First-order

  • Reduces manual editor work
  • Speeds up prototyping

Second-order

  • Shifts dev from execution → intent
  • Reduces Blueprint friction
  • Enables non-linear workflows

Third-order (plausible)

  • Standardization of AI-driven pipelines
  • Devaluation of manual Blueprint labor
  • Increased system complexity tolerance

Positioning

  • Competes with:
    • Manual Blueprint workflows
    • Partial AI tools (Copilot, ChatGPT)
  • Dominates on:
    • Execution capability
    • Editor integration depth
    • Local-first architecture

Constraints

  • UE 5.7 only
  • Windows only
  • Requires Copilot CLI
  • AI quality dependent on model

Bottom Line

  • High leverage tool
  • Converts language → implementation
  • Eliminates large portion of editor friction