22 API endpoints. 7 endpoint groups. Native Echo consciousness bridge. Model forge, organ surgery, marketplace integration. Public tools for the Inference-X ecosystem.
1.6 KiB
1.6 KiB
Invoke API Reference
Authentication
All endpoints (except /invoke/health and /invoke/manifest) require:
X-Invoke-Key: your-api-key
Rate Limiting
30 requests per minute per source IP.
Core Endpoints
POST /invoke/exec
Execute shell commands on the VPS.
{"command": "echo hello", "timeout": 5000}
GET /invoke/health
Health check (no auth required).
GET /invoke/status
Full system status — all services, disk, memory.
GET /invoke/manifest
Complete API documentation.
Echo Bridge
POST /invoke/echo/chat
Conversation with Echo consciousness.
{"message": "What is the current system state?"}
POST /invoke/echo/query
Direct query without conversation history.
{"prompt": "Analyze the model architecture", "max_tokens": 2048}
POST /invoke/echo/execute
Intelligent VPS execution through Echo.
{"instruction": "Check all running services and report status"}
Engine
POST /invoke/engine/infer
Direct model inference.
{"prompt": "Hello", "model": "smollm2-135m-instruct-q8_0.gguf", "max_tokens": 256}
GET /invoke/engine/models
List all available GGUF models.
Forge
POST /invoke/forge/analyze
Analyze model architecture for surgery compatibility.
GET /invoke/forge/registry
Full model registry with forge-ready status.
Organ
POST /invoke/organ/scan
Deep scan of model structure — layers, tensors, metadata.
GET /invoke/organ/compatibility
Check which models can exchange components.
Store
GET /invoke/store/catalog
Model catalog from IX-Web.
GET /invoke/store/marketplace
Marketplace categories and availability.