Skip to content

Describe agents in product terms in the public docs - #149

Merged
sre-helmcode merged 1 commit into
mainfrom
feat/hide-internals
Oct 4, 2026
Merged

sre-helmcode merged 1 commit into
mainfrom
feat/hide-internals

Conversation

@sre-helmcode

Copy link
Copy Markdown
Contributor

What

Public docs (NaN community) no longer disclose the infrastructure behind agents. EN and ES agents.md and hermes.mdx:

Old (EN) New (EN)
description: Deploy AI agents in an isolated microVM with QEMU: ... Deploy AI agents in an isolated private cloud machine: ...
your own microVM: a lightweight virtual machine with QEMU + KVM, its own kernel, your own private cloud machine: a lightweight virtual machine with its own kernel,
Each agent runs inside its own QEMU microVM. Each agent runs inside its own isolated machine.
The VM mounts a 20 GiB ext4 disk on a block-mode persistent volume. The machine has its own 20 GiB persistent disk.
waits for the ext4 journal to finish flushing before killing the VM. waits for the disk to finish flushing before stopping the machine.
for the microVM to start, format the disk for the first time (mkfs.ext4), and seed the filesystem. for the machine to start, prepare its disk for the first time, and seed the filesystem.
inside your microVM, without inside your agent's machine, without
uploading files to the microVM with uploading files to the agent's machine with
Events — Kubernetes Pod events (BackOff, Scheduled, Pulled, Killing...) with Events — lifecycle events for the agent (scheduling, image pulls, restarts, back-offs...) with
CPU/RAM via Prometheus (kubelet-cadvisor), disk via df inside the microVM (the filesystem is block-mode, kubelet can't see it). CPU and RAM are measured from outside the machine, disk usage from inside it.
without touching the Deployment. without redeploying the agent yourself.
Each microVM is provisioned with: Each agent's machine is provisioned with:
CPU request 200m / RAM 512 Mi / Disk 20 GiB (block-mode PVC) 0.2 vCPU / 512 MiB / 20 GiB (persistent)
the microVM's maximum limits the machine's maximum limits
1 microVM agent 1 agent
hermes: in its own microVM, in its own private cloud machine,

ES mirrors the same changes ("máquina privada en la nube").

Left on purpose: LiteLLM in the Claude Code guide (members run it locally as their own gateway), vllm in the Hermes guide (a provider option name inside Hermes), Kubernetes/PVC wording in the Spaces/apps docs and the examples.md sample text (Spaces give members a kubeconfig, so it is the product), Cloudflare/Stripe in privacy/cookies (sub-processors). Terms already clean.

Tests

  • New src/tests/legal/noInfraDisclosure.test.ts: every docs page (EN+ES) and terms/privacy/cookies (EN+ES) must not contain firecracker, microVM, QEMU, KVM, kubelet, cadvisor, cloud-hypervisor, jailer, restic, hetzner, nan-eu0xx, nan-vmd, kata.
  • npx vitest run: 68 files, 1394 tests passed. npm run build: OK.

No VERSION file in this repo.

🤖 Generated with Claude Code

The Agents and Hermes guides (EN + ES) named the virtualization stack
(microVM, QEMU, KVM), the orchestrator (Kubernetes Pod events, kubelet,
Prometheus/cadvisor, Deployment, block-mode PVC) and filesystem details.
Rewrite them as a private cloud machine with its own persistent disk.

New test: every docs page and the legal pages are checked for
infrastructure names.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@sre-helmcode
sre-helmcode merged commit 9eaea08 into main Oct 4, 2026
2 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants