Editorial policy

Source priority

1. Use official docs and release notes as the primary source for commands and product behavior.
2. Use GitHub issues and discussions for failure patterns, not as final truth.
3. Use community threads for user language and objections, then verify against primary sources.
4. Mark first-hand tests separately from inference.

Update workflow

• Review install, Windows, Docker, migration, security, skills, and provider docs weekly during the first 90 days.
• Update Last verified only after reviewing sources.
• Add a caveat when a page covers fast-moving behavior.
• Re-run route, metadata, sitemap, disclaimer, and accessibility smoke tests after material content updates.

Trust rules

• Do not claim official affiliation, endorsement, certification, or partnership unless documented.
• Disclose affiliate links before they are added.
• Do not publish real secrets, private prompts, or user data in examples.
• Prefer fewer useful pages over large batches of thin keyword variations.

People-first content standard

Every guide should exist because it helps an operator make a decision, complete a setup, avoid a risky default, or diagnose a failure. Search traffic is a distribution channel, not the reason a page exists.

When AI tools help draft content, the final page must still show human editorial choices: source priority, caveats, what was tested, what was inferred, and why the article is useful beyond summarizing public docs.

Structured data policy

• Use JSON-LD only when it represents visible page content.
• Do not add review ratings, product claims, FAQ schema, or affiliation claims unless the page visibly supports them.
• Keep breadcrumbs visible in the page and mirrored in BreadcrumbList JSON-LD.
• Prefer Article, CollectionPage, WebSite, Organization, and BreadcrumbList for this phase.

Direct answer

Agent Guide publishes independent operator guidance. Claims should be source-backed, dated, and honest about what was tested, inferred, or still unknown.

This page is part of Agent Guide's independent Hermes Agent cluster. It is source-backed and labelled when first-hand execution has not been run.

What this page does not cover

• It does not replace the official Hermes Agent documentation.
• It does not claim official affiliation with Nous Research.
• It does not claim first-hand testing unless the page explicitly says so.