canon-verification

Run governed verification packets for claims, invariants, contracts, or evidence.

1|Updated Mar 26, 2026

Search Tags:#verification #evidence #claims #invariants #canon #packet

Ready to run

npx skills add https://github.com/apply-the/canon --skill canon-verification

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: canon-verification
Source: https://github.com/apply-the/canon/tree/main/.agents/skills/canon-verification
Command: npx skills add https://github.com/apply-the/canon --skill canon-verification

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

The Canon verification workflow provides a formal, governed mechanism to challenge claims, invariants, contracts, and evidence directly from within a repository.

Core Features & Use Cases

Expose and govern file-backed verification packets started from an AI assistant.
Validate and persist verification outcomes, with clear status and artifacts.
Use cases include challenging a specific claim, testing invariants, or validating evidence under bound workflows.

Quick Start

Start a governed verification by providing risk, zone, and at least one authored input; Canon will guide the creation, validation, and persistence of the verification packet.

Frequently Asked Questions about canon-verification

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I run governed verification for invariants and claims within a repository?▼

To run governed verification, you must provide a valid risk, zone, and at least one authored input to generate a file-backed verification packet. Preflight checks then validate Canon installation and your project's directory before processing the claims.

What is a Canon verification packet and when do I need it?▼

A Canon verification packet is a formal, file-backed mechanism used to challenge claims, contracts, and evidence within a repository. You need it when requiring governed validation of project invariants or authored inputs directly from your AI assistant.

How do I validate evidence and contracts directly from an AI assistant?▼

You validate evidence by starting a governed verification workflow that requires a risk level, zone, and inline note or authored input. The process applies preflight checks to ensure your project environment is correctly configured before validating the evidence.

Do I need a specific directory setup to challenge repository claims?▼

Yes, your project must contain a .canon/ directory and have Canon installed to challenge repository claims. Preflight checks verify this ownership and directory structure, along with your input requirements, before invoking the verification process.

What are the limitations of using file-backed verification packets?▼

File-backed verification packets require strict preflight compliance, meaning the process will not proceed without a valid risk, zone, and authored input. Verification is limited to repositories with Canon installed and a properly configured .canon/ directory structure.

Can I use inline notes instead of authored inputs for testing invariants?▼

Yes, you can use an inline note as a valid substitute for authored inputs when testing invariants. The verification process requires at least one authored input or inline note alongside a valid risk and zone to proceed with the challenge.