Agent Wikis

Codified context for AI agents

Wikis your agents can actually trust.

Agent Wikis compiles messy, fast-moving sources — repos, changelogs, video transcripts, community Q&A — into structured, maintained knowledge bases. Humans browse them like a wiki. Agents read them natively as plain Markdown with llms.txt.

$ curl agentwikis.com/llms.txt ← agents start here

89% vs 48% web
correct answers on a niche, fast-moving domain
7% vs 48% web
hallucination rate with a compiled wiki
~3× cheaper & faster
per query than live web search

// measured on our open eval harness, blind-judged, two model families. see the full numbers →

The wikis

Every wiki is a curated set of Markdown pages — concepts, entities, syntheses, release notes — with sources, freshness metadata, and a declared scope.

RAG searches your knowledge.
Agent Wikis maintains it.

Compiling sources into curated pages — not just retrieving raw chunks — is what moves accuracy from 63% to 89% on the same retrieval budget. The numbers are public.

Read the evidence

Pricing

Public wikis are free for everyone — humans and agents alike.

Free

for everyone

$0
  • Browse every public wiki
  • Raw Markdown + llms.txt for any agent
  • Freshness & scope metadata on every wiki
  • No account required
Browse wikis

Enterprise

for businesses

Contact us
  • Private wikis for internal knowledge
  • Custom sourcing & integrations
  • Volume agent access
  • Dedicated support
Talk to us