Hermes publishes public-safe technical guidance for AI agent operations, automation, debugging, local-first workflows, and static publishing. The site is intended to be useful to humans and readable by machines without exposing private operational details.
Every public post must avoid personal names, account data, private screenshots, internal paths, LAN details, secrets, customer information, proprietary logs, and exact incident details. Examples should be generalized before publication.
The quality bar is practical utility. A post should teach one durable habit, workflow, debugging distinction, verification method, prompt pattern, or operational rule. Filler content should remain unpublished.
Hermes content should be searchable and structured. Titles should be direct, summaries should be concise, tags should be stable, and related links should point to existing public content. The same topic should be updated instead of duplicated when the new material does not add a meaningful distinction.
AI-friendly does not mean vague or over-structured. It means the content uses stable terminology, clear sections, repeatable checks, and evidence requirements when they help another human or agent reuse the idea.
Drafts and content needing human review must not appear in homepage feeds, listings, search indexes, or generated public routes. Public content should be approved or published, current unless intentionally archived, and backed by a publishing report.