Knowledge Policy
Unlike MCP, which is still planned, Knowledge Policy is real and effective today — it doesn’t depend on any access mechanism existing first. It’s this documentation site’s own stated rules for how an AI agent (or a person) should treat what it reads here: what to trust, how to judge whether something is current, how to cite it, and what to do when the answer isn’t actually here.
The Five Areas
Section titled “The Five Areas”- Canonical Sources — which pages are authoritative, and how this site marks the difference between real, published content and a page that isn’t written yet.
- Freshness — how to tell whether a page’s information is current.
- Citations — how to reference this site’s content correctly.
- Prohibited Assumptions — things never to assume about SOLMEME just because they’d be plausible.
- Uncertainty — what to do when this site doesn’t have a confirmed answer.
Where This Policy Comes From
Section titled “Where This Policy Comes From”This isn’t an aspirational policy written for AI agents in the abstract — it’s the same discipline this site’s own content is built under. Every real page on this site is produced through a documented authoring process that refuses to publish an invented fact, and every still-unwritten page is marked as such rather than silently filled with plausible-sounding text. The five pages in this section describe that same discipline from the reader’s side, for an agent instead of an author.
Where to Go From Here
Section titled “Where to Go From Here”- AI Tools Overview — how this fits into this site’s wider AI-access layer
- LLMs.txt — the real, live mechanism this policy applies to today