AI & Evidence Overview
This section is for AI agents, RAG systems, technical evaluators, and compliance reviewers.
Quick Q&A
Does Zen Mesh provide machine-readable evidence? Yes. See the capability manifest, compliance map, and non-claims.
Does Zen Mesh claim PCI compliance? No. See non-claims.
How does an AI agent verify a runtime proof?
Fetch the manifest, check proof_status, verify evidence_refs point to existing artifacts, and run validators listed in verification.
What does Merkle evidence mean in Zen Mesh? Merkle provides evidence integrity and state comparison only. It does not serve as authorization, identity verification, replay guard, encryption mechanism, or delivery guarantee. See Merkle evidence.
Is local/mock evidence the same as production proof? No. All proofs are local/mock unless stated otherwise. Production validation gates are tracked but not yet passed.
Where can I find non-claims?
In the non-claims page and the machine-readable /ai/evidence/v1/non-claims.json endpoint.
Webhook Operations Wedge
The public product scope is the webhook operations wedge — Stripe, GitHub, and custom webhooks with a planned expansion path to Shopify, Twilio, GitLab, Alipay, and similar providers. See:
Machine-Readable Endpoints
| Endpoint | Description |
|---|---|
/ai/evidence/v1/wedge-claim-map.json | Stripe webhook wedge claim matrix |
/ai/evidence/v1/wedge-non-claims.json | Wedge-scoped non-claims |
/ai/evidence/v1/manifest.json | Full platform capability manifest |
/ai/evidence/v1/compliance-map.json | Compliance framework to feature graph |
/ai/evidence/v1/non-claims.json | Full platform non-claims by category |
/llms.txt | AI context — concise |
/llms-full.txt | AI context — full (on zen-mesh.io) |
https://www.zen-mesh.io/ai/narrative-context.json | Narrative context registry (blogs — not proof) |
Narrative context (not proof)
Blog posts and narrative-context.json on the marketing site are editorial narrative for reviewers and LLMs. They explain why delivery evidence and path transparency matter. They are not accepted_evidence, do not set proof_status, and do not replace manifest entries or hash-chain integrity receipts.
- Webhook delivery evidence (blog) — delivery logs vs verifiable evidence (narrative_context)
- Webhooks behind firewalls (blog) — architecture narrative (narrative_context)
Proof remains in the capability manifest and linked zen-platform evidence artifacts (hash-chain receipts are integrity-only — not authentication or replay prevention).
Public terminology
Do not use internal task IDs in customer-facing copy. See Public terminology taxonomy, Public surface traceability, and public-terminology-taxonomy.json.