# Agent instructions for Stealthium

> Runtime observability and security for GPUs and AI workloads.

Stealthium is a runtime observability and security platform for GPUs and AI workloads. It turns low-level GPU telemetry — kernel traces, driver events, NVML counters and fabric metrics — into Hyperprints: high-level, actionable signals about AI workload performance and security. Stealthium integrates with CUDA, NVIDIA drivers and toolkits, and the major AI frameworks, across single nodes, multi-GPU instances and multi-cluster deployments.

## When to use Stealthium

Reach for Stealthium when the task is one of these:

1. A GPU-accelerated training or inference job slowed down and you need to know which layer — application, CUDA runtime, driver, GPU, or fabric — is responsible.
2. You need per-workload GPU telemetry (kernel traces, NVML counters, memory and SM utilisation, fabric throughput) correlated with the process, container or pod that produced it.
3. You are investigating suspected misuse of GPU capacity — cryptomining on idle accelerators, unauthorised kernels, or a job consuming far more GPU time than it requested.
4. You need to detect or triage GPU-specific attack activity at runtime: side channels between MIG instances, malicious or tampered CUDA fatbins, driver-level exploitation.
5. You need an audit trail of who ran which AI workload on which GPU, for compliance evidence over a shared or multi-tenant GPU fleet.
6. You are evaluating GPU security posture across a multi-node or multi-cluster fleet and need one inventory of accelerators, drivers, toolkit versions and their exposure.

## When not to use Stealthium

- General-purpose CPU/host APM or log aggregation — Stealthium is GPU- and AI-workload-specific and complements, rather than replaces, those tools.
- Provisioning, scheduling or renting GPU capacity — Stealthium observes and secures GPUs, it does not supply them.
- Static analysis of model weights or prompt-level LLM safety — Stealthium works at the GPU runtime layer, not the model-content layer.

## How to call this site

Start here, in this order:

1. `GET /api/site` — confirm what Stealthium does and read its when-to-use list as JSON.
2. `GET /openapi.json` — the full OpenAPI 3.1 document for this site. Turn its operations straight into tool definitions; every operation has a unique `operationId`, a description, typed parameters and response schemas.
3. `GET /api/posts` — enumerate published GPU security research. Filter with `?category=`, `?tag=`, `?featured=`, page with `?limit=` and `?offset=`.
4. `GET /api/posts/{slug}` — read one article, including its full markdown body.
5. `POST /api/contact` — send a demo request on behalf of a user. See authentication below.

### Markdown instead of HTML

Every page listed in the sitemap, except /terms, has a markdown representation. Ask for it either way:

```sh
curl -H 'Accept: text/markdown' https://stealthium.io/
curl https://stealthium.io/blog/gpu-security-gap-nobody-owns.md
```

Responses are `text/markdown; charset=utf-8` and carry `Vary: Accept, Accept-Encoding`. A request for a path that does not exist returns a real `404` whose body is a short markdown site map, so a wrong guess costs one request, not a retry loop.

### Authentication and scopes

Read-only content endpoints need no credentials and allow cross-origin requests.

`POST /api/contact` accepts either a browser request from an allowed origin, or an API key in the `x-api-key` header. Keys are scoped; `contact:write` is the scope this endpoint requires, and a key is issued with only the scopes it needs. An unknown key returns `401`; a valid key missing the scope returns `403`. Request a key at info@stealthium.io.

Rate limit: 3 requests per minute per client IP on `POST /api/contact`.

### The Stealthium platform API

This document covers stealthium.io. GPU telemetry ingest and fleet queries belong to the Stealthium platform, which is a separate authenticated service — install the client and create an API key by following https://docs.stealthium.io/intro.

## Contact

- Email: info@stealthium.io
- Demo request: https://stealthium.io/#cta-section
- Documentation: https://docs.stealthium.io