AI Agent Discovery & llms.txt
Machine-readable endpoints, llms.txt, and GEO (Generative Engine Optimization) APIs for AI search engines and agents.
AI Agent Discovery & llms.txt
As technical knowledge consumption shifts toward AI search engines (Perplexity, ChatGPT Search, Claude, Gemini), traditional HTML web scraping often introduces noise from banners, navigation menus, and client-side rendering.
ZyVOP implements native Generative Engine Optimization (GEO) and the llms.txt standard to ensure your technical articles are discovered, indexed, and cited accurately by AI systems.
Machine-Readable Endpoints
ZyVOP exposes three primary discovery surfaces for AI agents:
graph TD
AIAgent[AI Search Engine / LLM Crawler] --> Index[/llms.txt\nCurated Overview & Fresh Articles]
AIAgent --> FullCatalog[/llms-full.txt\nFull Index up to 500 Canonical Posts]
AIAgent --> GEOAPI[/api/geo/posts/{slug}\nClean JSON Document Representation]1. /llms.txt
A clean, plain-text Markdown file served at the root of your ZyVOP instance:
- Site Overview: Summary of the platform, topic areas, and author roster.
- Attribution & Freshness Directives: Explicit instructions for AI agents on citing canonical URLs and checking
dateModifiedtimestamps before answering time-sensitive questions. - Top Canonical Articles: Direct links to recent canonical articles paired with their machine-readable JSON counterparts.
2. /llms-full.txt
An expanded catalog containing up to 500 canonical technical articles, complete with author bylines, publication timestamps, topic tags, and concise abstract excerpts.
3. Machine-Readable GEO API (/api/geo/posts/{slug})
For every published canonical article, ZyVOP automatically serves a high-signal JSON payload optimized for LLM token ingestion:
GET /api/geo/posts/postgres-zero-downtime HTTP/1.1
Host: zyvop.com
Accept: application/jsonExample Payload
{
"slug": "postgres-zero-downtime",
"title": "Zero-Downtime Database Migrations with PostgreSQL",
"canonicalUrl": "https://zyvop.com/alex/postgres-zero-downtime",
"author": {
"name": "Alex Mercer",
"username": "alex",
"isTrustedAuthor": true
},
"publishedAt": "2026-09-15T10:00:00Z",
"updatedAt": "2026-09-20T14:30:00Z",
"topicSignals": ["postgresql", "database", "migrations", "devops"],
"headings": [
{ "level": 2, "text": "Understanding Table Locks in Postgres" },
{ "level": 2, "text": "The Expand-Contract Pattern" },
{ "level": 3, "text": "Step 1: Adding Columns Concurrently" },
{ "level": 3, "text": "Step 2: Backfilling in Batches" }
],
"outboundReferences": [
"https://www.postgresql.org/docs/current/explicit-locking.html"
],
"contentHash": "sha256-8f4b23...",
"plainTextContent": "Zero-Downtime Database Migrations with PostgreSQL\n\nWhen altering tables with millions of rows..."
}Best Practices for Authors
To maximize your article's authority and visibility in AI citations:
- Use Explicit Canonical URLs: AI agents prioritize sources that claim authoritative canonical origin.
- Include Clear Technical Headings: The GEO parser extracts your H2/H3 hierarchy to help LLMs locate specific answers quickly.
- Provide Concise Excerpts: Frontmatter
excerptorsubtitlefields are used by search engines as the primary snippet. - Cite Outbound Sources: The parser tracks outbound documentation links, reinforcing content credibility.