Skip to content

Operations

The Smart LLM Supervisor Gateway has two audiences for its documentation:

  • Users and integrators — covered by this public site. They use the gateway through the OpenAI-compatible API and do not need shell access to the production VPS.

  • Operators — covered by docs-internal/ in the repository. They have shell access to the production VPS and are responsible for deployment, backup, restore, monitoring, and incident response.

This page explains what is in docs-internal/, why the public site does not contain it, and how to request access.

Why split?

The public documentation site is published at docs.smart-openrounter.bee1x.one. It is reachable by anyone on the internet. Pages that contain server paths, SSH procedures, backup locations, or credential rotation steps must not appear there. A misconfigured indexer or a copy-paste into a public issue should not leak the path to the production PostgreSQL volume or the cert renewal hook.

The two-directory split (docs-public/ and docs-internal/) makes that boundary obvious in code review.

What is in docs-internal/?

Section Contents
docs-internal/operations/ Operator handoff, production topology, daily / weekly / monthly checklists, backup and restore, rollback, secret rotation, TLS certificates, incident response, provider outage handling, database recovery, disaster recovery.
docs-internal/development/ Local development, testing, repository layout, contributing, release process, routing change policy, model upgrade policy.
docs-internal/security/ Operational security, credential rotation history, audit trail.
docs-internal/evidence/ Index of evidence artifacts (with redaction policies).

How to access docs-internal/

docs-internal/ is part of the repository. Anyone with read access to the repository can read it. The repository is private in v1.0.0; access is by invitation from the maintainers.

The public site is built only from docs-public/. The MkDocs configuration (mkdocs.yml) explicitly points docs_dir at docs-public/.

How to request access

If you are taking over operations, follow the procedure in docs-internal/operations/operator-handoff.md:

  1. Confirm you have shell access to the production VPS.
  2. Confirm you have read access to the repository.
  3. Read the operator handoff document end-to-end.
  4. Run the day-1 checklist.
  5. Schedule a handover call with the previous operator.

If you are an end user

You are in the right place. The public documentation site is designed to give you everything you need to integrate, debug, and use the gateway. If something is missing, open an issue.

See also

  • DOCUMENTATION-HANDOFF.md at the repository root — how to maintain these docs (operator-only).
  • docs-internal/operations/operator-handoff.md — if you have access.