← The AI Hype Audit — all 296 verdicts
PARTLY Partly true: a 'semantic SEO wiki as a second brain for Claude, Codex, Gemini and Hermes' is a real and sensible pattern, but the post shows a graph screenshot and asks for comments instead of showing anything
The claimPhoto of a laptop showing a node-and-edge knowledge graph. Caption: 'Semantic SEO Wiki for my AI Agents. A 2nd Brain for Claude, Codex, Gemini, Hermes and more. Would you be interested on how to build your own Wiki lemme know in comments.' 194 reactions, 7 shares.
The idea is legitimate and increasingly standard: keep a structured, linked markdown knowledge base (an Obsidian-style vault, which is what the graph view in the photo is) that coding and research agents can read as shared context, so every agent starts from the same definitions instead of re-deriving them. For an SEO practice that means a wiki of entities, topic clusters and internal-linking rules the agents consult before writing. That is a real workflow and the poster is a real SEO consultant with a public practice. What the post contains, though, is a photo of a graph and a request for comments. No structure, no example page, no explanation of how the agents read it (a folder, an MCP connector, a retrieval index), and no result attributed to it. A commenter on the post says it plainly: the post collects comments rather than sharing knowledge. That is the mechanic here, an 'interested? comment below' engagement loop that may or may not ever produce the promised how-to. Nothing false is claimed, because almost nothing is claimed.
What holds up
- Post read via logged-in Chrome (caption, image and comment thread): a graph-view screenshot on a laptop, 194 reactions, and a 'let me know in comments' ask. Replies include one describing the post as comment collection.
- The 'second brain for agents' pattern (a semantically organised markdown knowledge base agents retrieve from) is documented widely in 2026 agent-workflow guides and is the intended use of vault-style tools with agent connectors.
- The poster runs a public SEO consultancy specialising in semantic and local SEO; the wiki concept fits that practice. No method, page, or result is shown in the post.
- No paid product is named in the post; the ask is engagement.
What doesn’t
- The post shows a screenshot and asks for comments; it does not show how the wiki is built, how agents read it, or what it changed.
- 'Interested? let me know in comments' is an engagement mechanic; the how-to is a promise, not a deliverable.
- 'Hermes' and 'Codex' are name-dropped as consumers of the wiki with no indication of how any of them connects to it.
The catch
The pattern is worth stealing; the post is not worth commenting on. Build a small markdown wiki of your own terms, rules and examples, point your agents at the folder, and you have the thing the screenshot implies, without waiting for a how-to that may never arrive.
How to actually do it
- Start a folder of plain markdown files: one page per core concept, product, audience and rule, with links between them. That is the whole 'wiki'.
- Give your agent read access to the folder (a project folder in Claude, a connector, or a retrieval index) and instruct it to consult the wiki before writing.
- Add a page each time an agent gets something wrong; the wiki grows from corrections, which is where its value comes from.
- Skip the comment. If the poster publishes the method, it will be findable; if not, you have already built yours.
A real, sensible pattern (a linked markdown knowledge base that agents read as shared context) presented as a graph screenshot plus a comment-harvest ask, with no method or result shown.
- Confidence
- High
- Posted by
- A verified SEO consultant's page (photo post sent to the self-thread Thu Sep 10, 2026, 7:22 AM)
We test hype for free. We build the real thing for a living.
Thirty minutes, no pitch — and you'll leave with something useful either way.
Book a call with Todd or start with the free Business Checkup →The Verdict Weekly
Three verdicts every Friday. Free forever, unsubscribe anytime, no spam — that would be ironic.
© Schreier Group · schreiergroup.com · See a wild AI claim? Drop it here and we'll test it.