Pro Sells Time, Not Safety — ELY Agent Input Preflight v0.8.0

ELY Agent Input Preflight v0.8.0 is released.

This release introduces an optional paid Pro entitlement, but I made one rule non-negotiable:

Pro sells time, not safety.

The Community edition keeps the complete verification and safety core.

GitHub Release:
https://github.com/Golovenkov79/ely-agent-input-preflight/releases/tag/v0.8.0

Product page:
https://elyzorix.ru/tools/agent-input-preflight/

What is ELY Agent Input Preflight?

ELY Agent Input Preflight is a local-first verification layer for AI coding agents.

Its job is not to replace the coding agent.

Its job is to independently answer questions such as:

• Is the project ready before the agent starts?
• Did the agent change files it was not supposed to change?
• Did generated Python code remain structurally valid?
• Did the requested work actually happen?
• Is there independent execution evidence?
• Can the final result be trusted instead of trusting only exit code 0?

The project is built around an important distinction:

PROCESS → EXECUTION → ACCEPTANCE → FINAL

A successful process exit is not automatically proof of a successful result.

What changed in v0.8.0?

v0.8.0 adds the foundation for an optional Pro entitlement.

New functionality includes:

• ely-preflight upgrade
• ely-preflight license status
• online Pro activation
• local license deactivation
• Ed25519-signed entitlement verification
• pinned public verification key
• local signed entitlement storage
• safe fallback to Community mode
• public license API via elyzorix.ru
• dedicated HTTP User-Agent for Cloudflare compatibility
• public Pro checkout and license claim flow

The important part is what did not change.

Community is still complete

I did not move the safety core behind the paywall.

Community still includes:

• Preflight checks
• project detection
• skills
• read-only agent mode by default
• execution evidence
• failure diagnosis
• diagnostic probes
• Generated-Code Admission
• Independent Postflight
• PROCESS / EXECUTION / ACCEPTANCE / FINAL semantics
• local run history
• core safety policy

You do not need Pro to know whether an AI coding agent did something unsafe or failed verification.

That would defeat the purpose of the tool.

So what is Pro for?

The design principle is simple:

Community protects correctness. Pro should reduce time and friction.

The paid layer is intended for convenience, automation and productivity rather than making the free version artificially unsafe or incomplete.

I think developer tools should earn money by making work faster — not by intentionally weakening the free safety layer.

Real-money end-to-end test

Before releasing v0.8.0, I tested the commercial route with an actual 499 RUB subscription purchase.

The verified path was:

payment
→ payment provider webhook
→ license issued
→ license claim
→ Windows activation
→ signed entitlement
→ PRO ACTIVE