Console

@cognipeer/console · Flagship · Control plane

stable

Self-hosted AI infrastructure with OpenAI-compatible API, provider routing & resiliency, vector orchestration, files pipeline, guardrails, tracing & observability, and project-level resource management.

Quickstart GitHub Full docs

License · MITTypeScript · 100%Audience · Platform teams that…Fits · Self-hosted AI platform

Self-hosted AI platformOpenAI-compatible APIsMulti-tenant operationsBrowse recipes

Production install

$npm install @cognipeer/console

What's in the box

Platform teams that need a multi-tenant AI gateway, control plane, and operational surface for inference, vectors, files, and tracing. Each capability is opt-in — use the parts that fit, leave the rest.

OpenAI Compatible API

Enables existing clients to operate as a seamless drop-in replacement, supporting chat, embeddings, streaming, and tool calling.

Provider Routing & Resiliency

Routes requests across model providers with built-in fallback handling and health-based routing strategies.

Vector Orchestration

Manages multiple vector databases through a unified control layer, including index lifecycle and operational workflows.

Files Pipeline

Supports document uploads with optional Markdown conversion, preparing content for RAG-ready ingestion pipelines.

Guardrails & Safety

Evaluate content against configurable safety policies with input and output filtering for production AI deployments.

Tracing & Observability

Tracks agent executions, tool calls, and LLM requests with integrated platform-level visibility.

Projects & Workspaces

Organise resources, models, and API keys into isolated projects with independent configuration and usage tracking.

How it runs

Console is the multi-tenant control plane: an OpenAI-compatible gateway sitting in front of providers, vectors, files, and tracing — with per-project quotas and guardrails baked in.

Client

Your app

OpenAI-compat

Layer 1

Gateway

Routing · quotas

Layer 2

Guardrails

In/out filters

Layer 3

Provider

OpenAI · Anthropic …

Layer 4

Vectors

Files · embeddings

Sink

Tracing

Audit · evals

Inputs

Chat completions
Embeddings
File uploads
Vector queries

Controls

Per-project quotas
Provider fallback
Guardrail policies
API key scopes

Outputs

Trace events
Token accounting
Audit log
Usage exports

Quickstart

Install, configure, run. The example below is the smallest piece of code that does something useful in production.

# Self-host the gateway with one command
$ docker run -p 8080:8080 ghcr.io/cognipeer/console:latest
 
# Call it from any OpenAI-compatible client
const res = await fetch("http://localhost:8080/v1/chat/completions", {
  method: "POST",
  headers: { "Authorization": "Bearer $COGNIPEER_KEY" },
  body: JSON.stringify({
    model: "gpt-4.1",
    messages: [{ role: "user", content: "Hi" }],
  }),
});

How it compares

Against the flagship · control plane options teams most often weigh — focused on operational concerns, not feature inventories.

Capability	Console	LiteLLM	Helicone	Portkey
Self-Hosted	● native	● native	● native	◐ partial
Open Source	● native	● native	● native	◐ partial
OpenAI Compatible	● native	● native	● native	● native
Routing / Fallback	● native	● native	◐ partial	● native
Rate Limiting	● native	● native	◐ partial	● native
Caching	● native	◐ partial	◐ partial	● native