AXIOM-MESH for AI agents and agent builders

Capability is not authority. Discovery is not permission. Connection is not permission.

AXIOM-MESH is a local-first coordination, authority, and evidence substrate. The current supported build is 0.12.0-dev.3, a production candidate, not production-promoted. This discovery surface is descriptive. Reading it, reaching the repository, finding a skill, or possessing credentials does not create permission.

Authority path

For supported privileged effects, the intended path is Gateway -> Hypervisor -> Sandbox -> Grid. A runtime, adapter, protocol, tool, or discovered capability must not become a parallel authority system.

Start with current truth

• README and current build claims
• Agent entry point
• Agent instructions
• Capability registry
• Current threat model
• Production readiness tracker

Portable authority audit

The AXIOM Authority Auditor is a self-contained, read-only Agent Skills-format procedure for separating capability from authority before consequential action. It does not execute actions, grant permission, replace policy, or certify a runtime.

Important non-claims

This page does not claim a live public AXIOM runtime, production promotion, completed independent security approval, production certification of external runtimes, a production MCP or A2A endpoint, BFT consensus, general remote execution, merge authority, or proof that signed evidence makes arbitrary external-world claims true.

Terminology

See the glossary for capability, authority, legitimacy, evidence, revocation, appeal, and meaningful exit.

Machine-readable navigation

• llms.txt
• llms-full.txt
• sitemap.md
• sitemap.xml
• Agent Skills discovery index
• Markdown mirror of this page

Publication status

The repository build artifact is prepared, not published until a separate deployment action is authorized and completed. A scanner score is diagnostic evidence about machine readability, not a security certification or authority grant.