Machine-facing intelligence
Give your AI a Witness.
WITNESS is built as an evidence and intelligence layer beneath the website, with machine-readable discovery and transport surfaces for compatible AI clients.
Canonical discovery
Entity
https://witnessstandard.com/entity.jsonhttps://witnessstandard.com/.well-known/witness.jsonhttps://witnessstandard.com/knowledge.jsonLLM source map
https://witnessstandard.com/llms.txtThe LLM source map identifies authoritative WITNESS public sources and the product boundaries that should be preserved when summarizing or classifying WITNESS.
MCP
Model Context Protocol discovery.
https://witnessstandard.com/.well-known/mcp.jsonMCP is a first-class surface for compatible clients that need to discover WITNESS capabilities without treating a website UI as the only control surface.
A2A
Agent-to-Agent evidence transport.
https://witnessstandard.com/.well-known/agent-card.jsonhttps://witnessstandard.com/a2a/extensions/witness-evidence/v1The public WITNESS A2A release artifact declares a core boundary: A2A is transport, not truth authority. Claims can move between systems without the transport layer itself deciding what is established truth.
Epistemic requirements
Machine-readable does not mean machine-established.
- Unknown remains a valid result.
- Model output cannot directly establish truth.
- Source reports, WITNESS analysis, and canonical evaluations remain distinguishable.
- Competing explanations and material uncertainty stay visible.
- Evidence quality can constrain downstream reasoning through Signal Integrity.
Public feeds